skills/dcf-model-builder/references/output-and-review.md

# Output and Review

Final response order:
1. Status and artifact mode.
2. Valuation range and scenario table.
3. Base-case bridge and key assumptions.
4. Sensitivities and what drives value.
5. Calculation integrity, decision readiness, hard failures, and warnings.
6. Source caveats, missing data, and next diligence.
7. Artifact paths.

Tone: direct, banker-ready, and caveated. Avoid false precision and do not bury screen-grade status below a valuation table.

Strong comments link value to drivers: growth, margins, reinvestment, WACC, terminal value, bridge, and source quality. Weak comments merely restate the formula or say the model "looks reasonable" without evidence.

On the first visible workbook tab and in the cover note, show two explicitly labeled status fields:
- `Calculation integrity`: formula, directionality, tie-out, and error-scan results.
- `Decision readiness`: source support, unresolved banker estimates, missing diligence, and permitted circulation posture.

Do not show one unqualified `Model checks: OK` status when decision readiness is `OPEN`, `partial`, or `screen-grade`.

For a public-company strategic alternatives pitch, address current trading, premium or discount read-through, what market expectations the base DCF implies, high terminal-value concentration, and material equity-bridge judgments such as leases or diluted-share treatment. When the base value shows a material premium or discount to trading, include a compact controlled numeric reverse-DCF read-through: show the market reference, base value gap, the assumption solved for, and the base assumptions held constant. Do not infer market-implied expectations only from a composite downside or upside case.

Keep scenario tables and discussion in a consistent order, ordinarily `Downside`, `Base`, `Upside`. Label quoted sensitivity amounts as a base-centered control range or the full displayed sensitivity spread so a wide displayed grid is not mistaken for the narrower calculation control.

Use `decision-grade` only when sources are current, bridge data is supported, checks pass, and assumptions are appropriate for the stated decision.