Skip to content

research-to-decision-memo.md

Latest commit

 

History

History
77 lines (52 loc) · 2.69 KB

research-to-decision-memo.md

File metadata and controls

77 lines (52 loc) · 2.69 KB
Raw

Research To Decision Memo

Who this is for: Analysts, PMs, operators, founders, consultants, and managers turning research into a decision.

Work problem: Research notes often become a pile of links and observations. This workflow turns them into a decision memo with options, tradeoffs, and a clear source boundary.

Inputs

What you need before running this:

  • Research question or decision to make.
  • Source notes, links, excerpts, or attached files.
  • Decision criteria and constraints.
  • Audience and desired memo length.

Copy-ready Codex instruction

Create a decision memo from the research inputs I provide.

Use only the sources, notes, and files I give you. Do not invent facts, statistics, customer quotes, market data, pricing, or legal conclusions. If a claim is not supported by the input, flag it as unsupported.

Decision question:
[write the decision]

Audience:
[who will read this]

Decision criteria:
[list criteria such as cost, speed, risk, quality, compliance, user impact]

Inputs:
[paste notes, source links, excerpts, or attach files]

Produce a decision memo with:
1. Executive summary: 5 to 7 bullets.
2. Decision question: one clear sentence.
3. Options considered: table with option, upside, downside, cost/risk notes, and source support.
4. Recommendation: one option, with reasoning.
5. Evidence map: claim, supporting source, and confidence.
6. Risks and unknowns: what could change the recommendation.
7. Human review needed: facts, numbers, assumptions, or approvals to check.

Keep unsupported claims out of the recommendation.

Expected output

A good run produces:

  • A memo that separates facts, judgment, and open questions.
  • A comparison table tied to stated criteria.
  • A recommendation with visible evidence and uncertainty.
  • A source map for important claims.

Human review checklist

Before sharing the memo:

  • Open the original sources for all important claims.
  • Check all numbers, dates, prices, and quoted facts.
  • Confirm the criteria match the real decision.
  • Add missing stakeholder, legal, security, or finance review where needed.
  • Decide whether the recommendation is supported enough to act on.

Failure modes

Where this workflow can go wrong:

  • It may over-weight sources with more detail.
  • It may treat weak evidence as enough for a recommendation.
  • It may hide disagreement between sources unless asked to surface it.
  • It may produce a polished memo that still lacks decision authority.

Source boundary

This workflow does not conduct independent research unless you explicitly ask Codex to browse or inspect sources. It does not verify facts beyond the material you provide. High-risk decisions need primary-source review.