financial-model-review

# Financial Model Review

## Problem

A model can look polished and still be dangerous. This skill reviews whether the
assumptions, structure, scenarios, and logic are strong enough to support a real
decision.

## Audience

- Primary: investors, diligence teams, and finance-heavy reviewers
- Secondary: operators reviewing planning or fundraising models

## When to Use

- Reviewing a startup, deal, or operating model before using it in a memo or decision
- Stress testing growth, margin, burn, or pricing assumptions
- Looking for missing scenarios, broken logic, or false precision
- Translating a spreadsheet into a model review memo with clear verdicts

## When Not to Use

- Building a model from scratch
- Market or competitor research without a model artifact
- Drafting the final memo when the model review is only one input: use `deal-memo-drafting`
- Synthesizing many mixed source documents: use `research-synthesis`

## Required Context

Gather or confirm:

- the model itself, or a faithful export of its assumptions and outputs
- what decision the model supports
- the business model and revenue engine
- the main value drivers or operating levers
- the time horizon
- whether the user wants an investor or operator framing

Useful but optional:

- prior versions, management guidance, or historical notes from Open Brain

## Process

1. Frame the review.
   - State the decision the model is being used for.
   - State whether the standard is investor-grade, board-grade, or internal planning.
2. Identify the model shape.
   - Revenue model, cost structure, cash runway, valuation, scenario design, and outputs.
3. Review assumptions.
   - Look for unsupported growth, margin leaps, pricing optimism, CAC efficiency, churn stability, and timing shortcuts.
4. Review structure and logic.
   - Flag missing drivers, circular reasoning, hidden hard-codes, inconsistent periods, or unsupported roll-forwards.
5. Review scenarios.
   - Check whether the model includes downside cases, key sensitivities, and break conditions.
6. Convert the review into judgment.
   - Distinguish fatal issues, caution flags, and acceptable simplifications.
7. Optionally use Open Brain.
   - Search for prior model assumptions, earlier reviews, or management claims.
   - Capture the final review memo or most important flags after completion.

## Evidence and Judgment Rules

- Prefer model evidence, historicals, source data, and management guidance over opinion.
- Do not pretend to verify formulas you cannot see. Say what is visible and what is not.
- Label unsupported assumptions as unsupported, not wrong by default.
- Call out where a model is useful for direction but not defensible for high-stakes precision.
- Always note missing scenarios if the downside case is absent or weak.
- Separate structural risk from business risk.

## Output

Default output:

- review objective and overall verdict
- key assumptions under pressure
- structural and scenario red flags
- what the model is good enough for
- what must change before the model supports a stronger decision

## Works Well With

- `competitive-analysis` when market benchmarks should inform assumption realism
- `research-synthesis` when the review needs source-backed contradiction handling
- `deal-memo-drafting` when the final deliverable needs an economics or risk section

## Notes

- This skill reviews what exists. It should not quietly turn into model building.
- The best outcome is a more decision-useful model, not a longer spreadsheet critique.