Proposal Skeleton Skill
Internal proposals die of three causes: the problem arrived unpriced (so the status quo stayed free), the options section was a strawman parade (so the reader felt managed), or the ask was unsized (so "interesting" substituted for a decision). The working spine is old and undefeated: the problem with its cost → real options fairly stated → the recommendation with reasons → risks conceded with mitigations → a decision-sized ask with a reversibility line. One page, because proposals are read in the gap between meetings.
What This Skill Produces
- The filled skeleton — problem-with-cost, options, recommendation, risks, ask — from the actual case
- The objections table — the three likely objections, pre-answered in the reader's dialect
- The ask, decision-sized — what exactly, costing what, decided by whom, reversible how
- The one-page version — with the appendix rule for everything that overflows
Required Inputs
Ask for these if not provided:
- The change and the evidence — what's being proposed and what supports it (the pilot data, the incident, the quote comparisons); skeletons organize evidence, not enthusiasm
- The cost of the status quo — the number or consequence that makes "do nothing" a choice with a price; without it, inertia wins by default and deserves to
- The decision-maker and their dialect — who says yes, and what they weigh (money? risk? team health?) — the recommendation argues in their currency (executive-summary audience rules apply)
- The honest alternatives — including the strongest version of "do nothing" and the rival option a smart skeptic would raise
Framework: The Spine Rules
- Price the problem first: "we spend ~11 hours/week on manual reconciliation (≈ $2.9k/month)" converts the proposal from a preference into arithmetic — and do-nothing from the safe default into the most expensive option on the page. Honest numbers only; one inflated cost discounts the whole case.
- Options are steelmanned or they're theater: 2–3 real options, each with its genuine best case — including do-nothing's ("zero switching cost, no adoption risk"). Readers detect strawmen instantly, and a fairly-stated rival makes the recommendation more credible, not less.
- Recommend with reasons that touch the comparison: "Option B, because it fixes the cost at 40% of Option C's price and reverses in a month if wrong" — the recommendation cites the dimensions the options table established, in the decision-maker's currency.
- Concede risks before the reader finds them: the objections table — likely objection → honest answer or mitigation — signals the thinking survived contact with skepticism. A proposal with no stated risks reads as either naive or salesy; both lose.
- The ask is decision-sized and reversibility-framed: "Approve a 6-week pilot for one team at $X, success = [metric], full rollout decided after" — small enough to say yes to today, with the exit named. Big irreversible asks get deferred forever; pilots get approved this week. The one-page rule: spine on page one, everything else (the vendor matrix, the data) appendixed and referenced.
Output Format
Proposal: [the change] — for [decision-maker]
The Problem, Priced
[Two sentences + the honest number/consequence of continuing]
Options (steelmanned)
| Option |
Its best case |
Cost |
Risk |
| [Do-nothing included, fairly] |
|
|
|
Recommendation
[The option + the reasons in the reader's currency + the reversibility line]
Objections, Pre-Handled
| They'll ask |
The honest answer |
The Ask
[Exactly what · costing what · decided by whom, by when · pilot-sized with the success metric and exit]
(Appendix: the overflow — referenced, not inlined)
Quality Checks
Anti-Patterns
1---2name: proposal-skeleton3description: Structure an internal proposal that gets a decision — the problem-cost-options-recommendation-ask spine, the objection pre-handling that shortens the meeting, and the reversibility framing that makes yes easier. Use when asked write a proposal for the new tool or process or hire, how do I pitch this internally, structure my case for the change, or my proposals keep dying in review. Produces the proposal skeleton filled from the actual case, the objections table, the decision-sized ask, and the one-page discipline.4---5
6# Proposal Skeleton Skill
7
8Internal proposals die of three causes: the problem arrived unpriced (so the status quo stayed free), the options section was a strawman parade (so the reader felt managed), or the ask was unsized (so "interesting" substituted for a decision). The working spine is old and undefeated: the problem *with its cost* → real options fairly stated → the recommendation with reasons → risks conceded with mitigations → a decision-sized ask with a reversibility line. One page, because proposals are read in the gap between meetings.
9
10## What This Skill Produces
11
12- **The filled skeleton** — problem-with-cost, options, recommendation, risks, ask — from the actual case
13- **The objections table** — the three likely objections, pre-answered in the reader's dialect
14- **The ask, decision-sized** — what exactly, costing what, decided by whom, reversible how
15- **The one-page version** — with the appendix rule for everything that overflows
16
17## Required Inputs
18
19Ask for these if not provided:
20- **The change and the evidence** — what's being proposed and what supports it (the pilot data, the incident, the quote comparisons); skeletons organize evidence, not enthusiasm
21- **The cost of the status quo** — the number or consequence that makes "do nothing" a choice with a price; without it, inertia wins by default and deserves to
22- **The decision-maker and their dialect** — who says yes, and what they weigh (money? risk? team health?) — the recommendation argues in their currency ([executive-summary](../executive-summary/SKILL.md) audience rules apply)
23- **The honest alternatives** — including the strongest version of "do nothing" and the rival option a smart skeptic would raise
24
25## Framework: The Spine Rules
26
271. **Price the problem first:** "we spend ~11 hours/week on manual reconciliation (≈ $2.9k/month)" converts the proposal from a preference into arithmetic — and do-nothing from the safe default into the most expensive option on the page. Honest numbers only; one inflated cost discounts the whole case.
282. **Options are steelmanned or they're theater:** 2–3 real options, each with its genuine best case — including do-nothing's ("zero switching cost, no adoption risk"). Readers detect strawmen instantly, and a fairly-stated rival makes the recommendation *more* credible, not less.
293. **Recommend with reasons that touch the comparison:** "Option B, because it fixes the cost at 40% of Option C's price and reverses in a month if wrong" — the recommendation cites the dimensions the options table established, in the decision-maker's currency.
304. **Concede risks before the reader finds them:** the objections table — likely objection → honest answer or mitigation — signals the thinking survived contact with skepticism. A proposal with no stated risks reads as either naive or salesy; both lose.
315. **The ask is decision-sized and reversibility-framed:** "Approve a 6-week pilot for one team at $X, success = [metric], full rollout decided after" — small enough to say yes to today, with the exit named. Big irreversible asks get deferred forever; pilots get approved this week. The one-page rule: spine on page one, everything else (the vendor matrix, the data) appendixed and referenced.
32
33## Output Format
34
35# Proposal: [the change] — for [decision-maker]
36
37## The Problem, Priced
38[Two sentences + the honest number/consequence of continuing]
39
40## Options (steelmanned)
41| Option | Its best case | Cost | Risk |
42|---|---|---|---|
43[Do-nothing included, fairly]
44
45## Recommendation
46[The option + the reasons in the reader's currency + the reversibility line]
47
48## Objections, Pre-Handled
49| They'll ask | The honest answer |
50|---|---|
51
52## The Ask
53[Exactly what · costing what · decided by whom, by when · pilot-sized with the success metric and exit]
54
55*(Appendix: the overflow — referenced, not inlined)*
56
57## Quality Checks
58
59- [ ] The status quo carries an honest price
60- [ ] Every option including do-nothing got its best case
61- [ ] The recommendation argues in the decision-maker's stated currency
62- [ ] Risks are conceded with mitigations before the reader raises them
63- [ ] The ask is pilot-sized, dated, and names its reversal path
64
65## Anti-Patterns
66
67- [ ] Do not propose against an unpriced status quo — free inertia beats every paid change
68- [ ] Do not strawman the alternatives — the detected strawman costs more credibility than the rival option ever would
69- [ ] Do not ask for the end-state — ask for the smallest decidable step toward it
70- [ ] Do not hide the risks — the reader's discovered objection beats your conceded one, in their favor
71- [ ] Do not exceed a page in the body — the appendix exists so the spine can breathe
gabrielmoreira