The Decision Triage Kit

The tier cheat sheet

Before you analyse any decision, size it. Money and time both count; whichever is larger wins.

Then ask two more questions:

Is it reversible? If you can undo it cheaply (a monthly subscription, a test campaign), bias toward action and set a review date. If it is hard to undo (a hire, a lease, a rebrand), slow down by one tier.

Is it actually urgent? Blocking revenue right now: decide within 48 hours. Strategic but not burning: take one to two weeks and do the full pass. Urgent but unimportant: delegate it or give it ten minutes. Neither: it is not a decision, it is a distraction. Park it.

The boundary rule: when the cost is ambiguous or the reversibility is unclear, treat it as the higher tier.

The reframe (do this before analysing anything above Micro): the question you asked is rarely the real one. “Should I buy X” is usually “what removes friction from [the outcome I want] in the next 90 days.” Write the real question down. Analyse that one.

The eight distortions

Scan your own reasoning for these before you decide. Naming the pattern is usually enough to correct for it.

1. Special money. Treating a windfall as different from operating cash.
2. Validation seeking. The numbers already answer the question; you want permission.
3. Comfort resistance. Rejecting the simple fix because it feels too easy.
4. Sunk cost. “I’ve already spent X, so I should…”
5. Future self bargaining. “I’ll do X once Y happens,” when Y is what X would enable.
6. Scarcity theater. Agonizing over small money while reserves sit adequate.
7. Analysis paralysis. Researching to avoid deciding. You have asked this question before.
8. Shiny object. The new option looks better than finishing the current path.

The decision record template

Copy this into a dated file (decisions/2026-08-12-hire-va.md) every time you make a Small decision or bigger. The last section is the whole point: on the review date, you fill in what actually happened and find out how good your judgment really is.

---
date: 2026-08-12
tier: small | medium | large
reversible: yes | no
confidence: 70
review-date: 2026-09-12   # +30 days if reversible, +90 if not
---

# Decision: [one line, the REAL question]

## The call
[Yes | No | Test-first | Defer] and one sentence.

## Why
The single biggest reason.

## Which means
The second-order effect: what this unlocks or compounds.

## What could go wrong
The top failure mode, and why I am proceeding anyway (or why it killed the idea).

## Distortion check
Which of the eight patterns showed up, if any, and what it was doing to my thinking.

## Implementation
1. First step, with a deadline:
2. Success looks like (a number):
3. Exit trigger (what makes me stop or reverse):

## Review (fill in on the review date)
- Expected outcome:
- Actual outcome:
- What I learned about my own judgment:

Make it automatic

This kit is the manual version. The full protocol runs as a /decide command in Claude Code: it triages, reframes, dispatches specialist agents on bigger decisions, catches the distortions for you, and writes this record automatically. The command is free in the toolbox.

Download →

← Back to the library