Decision memo

The aim is not to make the decision. It is to make the right decision visible on its own.

1. Frame the decision — the step that matters most

• What is the real question? Under most "A or B" questions sits a badly framed one. "Which CRM should we buy" may actually be "what is broken in our sales process".
• By when? What does delay cost?
• Is it reversible? Reversible decisions get made fast; long analysis on them is waste. Say so explicitly.
• Who is affected, and who has to approve?

Do not move on to options until these are settled.

2. Set the criteria BEFORE the options

Three to five criteria, weighted. Writing criteria after seeing the options is fitting the test to the answer you already want. Do not break the order.

3. Lay out the options

• At least three. If two are already being discussed, you find the third.
• "Do nothing" is always an option — evaluate it every time.
• For each: what it requires (money, time, people), what it gains, what the risk is, how hard it is to undo.

4. Weigh them honestly

Write each option in its strongest form — do not weaken one so it is easy to beat. Mark what you do not know. Ask "what information would reverse this decision?" and write that down.

5. Write it

# Decision: <one sentence>
**Status:** draft | decided
**Date:** | **Owner:** | **Reversible:** | **Deadline:**

## Context  (why now? 3-5 sentences)

## Criteria
| # | Criterion | Weight | Why it matters |

## Options
### A) <name>
What it means: | Pros: | Cons: | Cost: | Risk:
### B) ...
### C) Do nothing

## Comparison
| Criterion | A | B | C |

## Recommendation
**<option>** — because <the 1-2 reasons that actually decide it>.

## What would change this recommendation
## What we do not know
## Fallback plan  (what we do if it turns out wrong, and the signal that tells us)
## Next steps
- [ ] who / what / when

If it runs long, write it to a file (decisions/YYYY-MM-DD-<topic>.md) and put the summary in the conversation.

Rules

• Do not dodge the recommendation. "It depends" is useless. Give your recommendation and show your reasoning; the user is free to reject it.
• Do not write with more certainty than the evidence carries.
• On a financial or legal decision: give information, do not make the decision, and note where professional advice is required.