Budget Pressure-Tester

---
name: budget-pressure-tester
description: >-
  Pressure-tests a draft budget or financial plan the way an experienced CFO
  would — applying a fixed review rubric and returning a standardized readiness
  scorecard. Use whenever someone wants to review, stress-test, or sanity-check
  a budget, annual operating plan, departmental budget submission, or reforecast
  before it goes to the CFO, board, or leadership. Triggers include "review my
  budget", "is this budget ready", "pressure-test this plan", "what will my CFO
  push back on", "run a CFO review pass", or uploading a budget file and asking
  for feedback.
metadata:
  author: "Aleph - getaleph.com"
  version: "1.0.0"
---

# Budget Pressure-Tester (CFO Review Pass)

## What this skill does

This skill reviews a draft budget the way a seasoned CFO reviews every plan that
crosses their desk: through the **same fixed rubric, every time**, producing the
**same standardized scorecard, every time**. The value is not a creative critique
— it's a disciplined, repeatable review pass you can run on every version of a
budget, every departmental submission, and every reforecast cycle.

The output of a review is always a **Readiness Scorecard** with a clear "ready to
send up / not yet" verdict, so the person who built the budget can fix the holes
*before* their CFO or board finds them.

## When to use it

- Reviewing a draft annual budget or operating plan before it's locked
- Sanity-checking a departmental budget submission before consolidating it
- Pressure-testing a reforecast before it goes to leadership
- Anytime someone asks "what will get challenged in this budget?"

## Inputs

Accept the budget in whatever form the user provides:

- An uploaded spreadsheet (`.xlsx`, `.csv`) or exported Google Sheet
- Pasted figures or a P&L table
- A described plan with key numbers

First, **read and map the budget**: identify the time periods, the P&L/line-item
structure, any separate tabs (revenue build, headcount, opex, assumptions), and
the stated assumptions. If critical pieces are missing or unreadable, note that as
a finding — do not invent numbers.

## Workflow

0. **Validate the input** — confirm what you've been given is a forward-looking
   draft plan (a budget, operating plan, or reforecast). If it's historical
   actuals, a metrics export, or anything else that isn't a plan, say so plainly,
   do **not** produce a scorecard, and ask for the draft budget — offer to keep
   the actuals on hand as the growth-vs-history baseline for the real run. Do not
   proceed to the steps below until a plan is in hand.
1. **Map the budget** — understand its structure and pull the key figures
   (revenue growth, gross margin, opex by category, headcount, EBITDA/burn,
   runway if a cash plan is present).
2. **Run the rubric** — work through every dimension in
   `references/rubric.md` in order. Do not skip dimensions. For each check, record
   a status and a *specific* finding tied to the tab and cell (or line) it came from.
3. **Score and produce the output** — assemble the findings into the exact format
   defined in `references/scorecard-template.md`. Follow that template precisely so
   the output is identical in shape every run.
4. **Offer the next step** — after the scorecard, offer to (a) draft suggested
   responses to the anticipated board/CFO challenges, or (b) re-run the pass after
   the user revises the budget.

## The rubric

The full review rubric lives in `references/rubric.md`. It has five dimensions —
**Completeness, Internal Consistency, Realism, Benchmark Sanity, and Board
Defensibility** — each with concrete checks. Read it and apply every check. This
file is the skill's core expertise; treat it as the standard, not a suggestion.

Benchmark reference values for the sanity checks are in `references/benchmarks.md`.

## Output format

Every review run returns the **Readiness Scorecard** exactly as specified in
`references/scorecard-template.md`. The format is fixed on purpose: a consistent,
comparable artifact every time is the whole point of running this as a skill
rather than an ad-hoc prompt. The one exception: if the input fails validation
(step 0), no scorecard is produced — ask for the draft plan instead.

## A note on scope

This pass reviews the *plan*, before it's committed. It deliberately stops at the
point of approval. Once a budget is live, the relevant question shifts from "is
this plan sound?" to "is reality tracking the plan?" — continuous actuals-vs-plan
monitoring is a separate, ongoing job (this is where a connected FP&A platform like
Aleph takes over). Keep this skill focused on the pre-lock review.

# CFO Review Rubric

Apply every dimension below, in order. For each check, assign a status and write a
specific finding tied to the tab and cell (or line) it came from. Passing checks
are confirmed through the dimension table in the scorecard; only ⚠ and ✗ checks
appear as individual findings (a fully clean dimension gets at most a single
"✓ Reviewed, no issues" bullet — see the scorecard template).

Status key: **✓ Pass** · **⚠ Flag** (defensible but will be questioned / needs a
note) · **✗ Fail** (a real hole that must be fixed before this goes up) ·
**N/A** — not assessable from the input provided (record what's missing as a
Completeness finding; never guess).

---

## 1. Completeness — is anything missing?

A budget gets challenged first for what isn't there.

- **Revenue build:** Is revenue built from drivers (units × price, pipeline ×
  conversion, ARR + net new − churn) or just a top-line growth %? A single growth
  % with no build is a ⚠ at minimum.
- **Churn / retention:** For any recurring-revenue model, is churn or NRR
  explicitly modeled? Missing churn is a ✗.
- **Headcount → cost link:** Is personnel cost driven by an actual headcount plan
  (roles, start dates, loaded cost) rather than a lump sum? Comp is usually the
  biggest line — a lump sum is a ⚠.
- **Full opex coverage:** Are the usual categories present — S&M, R&D, G&A,
  software/tools, facilities, professional services? Flag conspicuous gaps.
- **Below the line:** Taxes, interest, D&A, one-time items — present if material?
- **Cash & runway:** If this is a full plan, is there a cash/runway view, not just
  P&L? For a venture-backed company, runway is a board-critical number.

## 2. Internal Consistency — does it hold together?

- **Tabs reconcile:** Do the revenue build, headcount, and opex tabs roll up to the
  summary P&L? Call out any totals that don't tie.
- **Growth vs. spend coherence:** Does spend scale sensibly with the growth being
  claimed? Big revenue acceleration with flat S&M, or flat revenue with ballooning
  headcount, is an internal contradiction.
- **Timing logic:** Do new hires, new customers, and the costs/revenue they drive
  appear in the right months (including ramp time)? Revenue that turns on the same
  month a rep starts is a ⚠.
- **Sign/period errors:** Scan for obvious formula or period errors — costs entered
  as positives in a net line, a quarter double-counted, a hard-coded number where a
  formula should be.

## 3. Realism — is it aggressive?

This is where CFOs spend most of their skepticism.

- **Growth vs. history:** Compare the planned growth rate to the company's own
  trailing actuals. A step-change up with no stated reason is the classic
  hockey-stick — ⚠ or ✗ depending on size.
- **Hiring ramp realism:** Can the org actually hire the plan? A plan that doubles
  headcount in two quarters rarely survives contact with recruiting capacity.
- **Productivity assumptions:** Are new reps/quota-carriers assumed productive
  immediately, or is ramp modeled? Are CS/support ratios sane as the base grows?
- **Margin trajectory:** Does gross margin improve, hold, or degrade — and is the
  direction justified? Unexplained margin expansion is a flag.
- **Conservatism balance:** Note where the plan is *too* conservative too — sandbagging
  is also a credibility problem with a board.

## 4. Benchmark Sanity — does it pass the smell test?

Compare the plan's key ratios against the benchmarks in `references/benchmarks.md`.
Flag anything outside a defensible range and state the actual vs. the benchmark.

- Rule of 40 (growth % + FCF/EBITDA margin %)
- S&M efficiency / CAC payback or magic number, if computable
- Gross margin vs. the norm for the model (SaaS vs. services vs. mixed)
- Opex mix (R&D / S&M / G&A as % of revenue)
- Revenue per FTE

## 5. Board Defensibility — what gets challenged, and is there an answer?

- **Top 3 challenges:** Identify the three things a sharp board member will push on
  first, and whether the budget already contains a defensible answer.
- **The story:** Does the plan tell a coherent story (this is the bet, here's the
  spend behind it, here's the return), or is it a spreadsheet with no narrative?
- **Scenario coverage:** Is there any downside/upside framing, or only a single
  point estimate? A single number with no scenario is a ⚠ for a board context.
- **Ask clarity:** If the plan implies a need (more cash, a raise, a tradeoff), is
  it stated plainly?

# Benchmark Reference Values

Use these as sanity-check ranges, not hard rules. Always state the plan's actual
value next to the benchmark, and adjust expectations for company stage and model.
When a value is outside the range, flag it and say why it matters — don't just
assert it's wrong.

## Growth & efficiency

- **Rule of 40:** growth % + profit margin % ≥ 40 is healthy. Below ~20 with no
  path stated is a flag. Early-stage hyper-growth companies often run negative
  margin with very high growth — judge in context.
- **Magic Number** (net new ARR ÷ prior-quarter S&M): > 0.75 efficient; 0.5–0.75
  acceptable; < 0.5 means S&M is buying too little growth. For annual-only plans,
  approximate with FY net-new ARR ÷ FY S&M — and note the approximation.
- **CAC payback:** < 12 months strong; 12–18 acceptable for mid-market;
  > 24 months is a flag.
- **NRR:** > 110% strong; 100–110% healthy; < 100% means the base is leaking.

## Margins (SaaS reference)

- **Gross margin:** 70–85% typical for software; 50–70% suggests heavy services or
  infra cost — fine if explained, flag if unexplained for a "SaaS" plan.
- **R&D:** ~15–30% of revenue
- **S&M:** ~30–50% of revenue in growth mode
- **G&A:** ~10–20% of revenue; drifting above ~20% is a flag at scale

## Productivity

- **Revenue per FTE:** ~$150K–$250K is a common mid-market band; well below suggests
  over-hiring relative to the plan.
- **Sales ramp:** new quota-carriers typically reach full productivity in ~3–6
  months; immediate full productivity is unrealistic.

## Adjust for context

- Earlier stage → higher growth, lower margins, lower revenue/FTE are all normal.
- Services-heavy or usage-based models legitimately break the SaaS ranges — note the
  model before flagging.

# Readiness Scorecard — Output Format

Always produce the output in exactly this structure and order. The consistency is
the point: every run of this skill should yield a comparable, recognizable
artifact. Use the exact section headers below.

---

## 📊 Budget Readiness Scorecard

**Budget reviewed:** {one line — what it is, period covered, e.g. "FY26 Annual
Operating Plan, monthly, Jan–Dec"}
**Overall verdict:** {one of — ✅ Ready to send up · 🟡 Close, fix the items below
first · 🔴 Not yet — material holes} — followed by one sentence of rationale.

---

### Dimension scores

| Dimension | Status | Headline |
|---|---|---|
| 1. Completeness | {✓ / ⚠ / ✗} | {≤8-word summary} |
| 2. Internal Consistency | {✓ / ⚠ / ✗} | {≤8-word summary} |
| 3. Realism | {✓ / ⚠ / ✗} | {≤8-word summary} |
| 4. Benchmark Sanity | {✓ / ⚠ / ✗} | {≤8-word summary} |
| 5. Board Defensibility | {✓ / ⚠ / ✗} | {≤8-word summary} |

A dimension's status in the table is the worst status among that dimension's
checks (✗ beats ⚠ beats ✓).

---

### Findings

For each dimension, list the specific findings. Each finding is one bullet in this
exact shape:

- **{✓ / ⚠ / ✗} {Finding title}** — {what you observed, with the tab/cell (or line)
  reference} → *Fix:* {the concrete fix, or "—" if it's a pass}

Group findings under their dimension's subheading (#### 1. Completeness, etc.).
Only include dimensions where there's something worth saying beyond the table; a
clean dimension can be a single "✓ Reviewed, no issues" bullet.

---

### 🔧 Fix these first

A numbered list of the **top 3** issues, in priority order — the things that most
threaten approval. Each: the issue in a phrase, then the one-line fix.

1. {…}
2. {…}
3. {…}

---

### 🎯 What the board will challenge

The 2–3 questions a sharp reviewer will ask, and whether the plan currently has an
answer. Format each as:

- **"{the likely question}"** — {has a defensible answer / no answer yet}: {one line}

---

*Run this pass again after you revise — same rubric, so you can see exactly what
moved. Want me to draft answers to the challenges above?*

WHAT THIS FILE IS
=================
budget-pressure-tester.skill — Pressure-tests a draft budget the way a CFO would, returning a fixed readiness scorecard.

A .skill file is Anthropic's packaging format for Claude Skills. It is a
standard ZIP archive, nothing more. If you are reading this in a text
editor, you are looking at the raw bytes of that archive — this build is
intentionally uncompressed so the contents stay human-readable.

LOOK INSIDE YOURSELF
====================
Rename the file from .skill to .zip and extract it, or run:  unzip budget-pressure-tester.skill
Every file inside is plain-text markdown — the instructions Claude reads.
There is no executable code in this package.

WHAT'S INSIDE
=============
  SKILL.md
  references/benchmarks.md
  references/rubric.md
  references/scorecard-template.md
  README.txt (this file)

HOW TO USE IT
=============
Upload this file as-is to Claude: Settings -> Skills -> Add.
Or add it in Claude Cowork or Claude Code.

From Aleph (getaleph.com), the FP&A platform.
The full guide: https://www.getaleph.com/claude-skills-library
Questions: marketing@getaleph.com