Board and management commentary with AI: draft the monthly narrative from tied-out numbers

When to use it

Use it once the numbers are final and the flux review is done. The drafting of "Revenue was $X, up Y% vs budget, driven by…" is time-consuming and repetitive, and AI drafts it well if every number comes from a source you've already reviewed. It drafts badly if you ask it to explain the business on its own.

Input data

• The final reporting package (P&L, balance sheet, cash flow, KPIs) for the period, with comparisons to budget/forecast and prior year.
• Reviewed flux or variance explanations (see the flux workflow).
• Last period's commentary as a style example.
• Key messages from the CFO, and anything that must not be said yet (pending deals, unannounced changes).

The prompt

Replace the [bracketed] items. Paste it with your files attached. It works in any capable assistant; tool-specific notes are in the steps.

Draft the [monthly management / quarterly board] financial commentary for [period].

Attached:
- Final reporting package: [file]
- Reviewed variance explanations: [file]
- Last period's commentary (match this structure, length and tone): [file]
- CFO key messages: [bullets]
- Do not mention: [list]

Rules:
1. Use only numbers that appear in the reporting package. After each number, add a bracketed source reference like [P&L!D14] or [KPI tab, row "ARR"]. I will strip these later.
2. Use only the reviewed variance explanations for "why". If a variance over [threshold] has no reviewed explanation, write "[EXPLANATION NEEDED]".
3. Compare to [budget/forecast] and [prior year] consistently. State $ and % changes; check your arithmetic and show it in a separate check table at the end.
4. Neutral, factual tone. No forward-looking statements or guidance unless they're in the CFO key messages verbatim.
5. Sections: headline (3 bullets), revenue, gross margin, operating expenses, EBITDA/operating income, cash and liquidity, KPIs, risks and watch items.

Steps

1. Run the prompt only after the reporting package is final and tied out.
2. Check every number against its source reference. The references make this fast, so don't skip them. Then check the arithmetic in the check table.
3. Fill the [EXPLANATION NEEDED] gaps from the business, not from the AI.
4. The CFO or controller edits for message and emphasis, then strip the source references.
5. For public companies, follow your disclosure controls: the commentary may be MNPI, and anything external goes through disclosure-committee review.

Control map

What the AI may do, step by step. How ratings work.

Audit trail

• Keep the version with source references alongside the final version, and the reporting package it was drafted from.
• Record who reviewed and approved the commentary before distribution.

Failure modes

• Arithmetic and percentage errors, especially percentage-point vs percent changes and sign errors on expense variances. Use the check table.
• Numbers from the wrong column (YTD vs month, budget vs forecast). The source references catch this.
• Invented causation ("driven by strong demand") where the reviewed explanation said something narrower.
• Leakage of sensitive items listed under "do not mention" through paraphrase. Read for it deliberately.

Data handling

• Use a company-approved AI workspace whose terms say your inputs are not used for model training. Check your plan; don't assume it.
• Strip or mask what the task doesn't need: employee names, bank account numbers, customer PII, SSNs or TINs.
• If you are a public company, treat pre-release results as material non-public information and keep them inside approved systems.