Finance Variance Narrative Generator
Variance tables tell you what changed. Executives want to know why it changed and what to do about it. This skill turns raw variance data into a clear, decision-oriented narrative.
When to use this skill
Use this skill when:
- You have actuals-vs-budget or actuals-vs-forecast variances to explain.
- You are preparing the commentary for a monthly or quarterly review.
- You want consistent, plain-language narratives instead of restating the numbers.
See references/variance-analysis-checklist.md for the review checklist and
templates/executive-narrative-template.md for the output structure.
Inputs needed
- The variance data: line item, actual, budget/forecast, and variance ($ and %).
- The period and the comparison basis (budget vs forecast vs prior year).
- Known drivers or one-time items, if any.
- The audience (board, CEO, department head).
Process
- Sort variances by absolute dollar impact and keep the material ones.
- Apply a materiality threshold so small noise is excluded.
- For each material variance, state the driver in business terms, not accounting terms.
- Separate one-time items from recurring trends.
- Note whether each variance is favorable or unfavorable to the bottom line.
- End with the "so what": the implication for the forecast or for action.
- Keep the narrative to the few variances that actually matter.
Prompt or workflow
You are writing executive variance commentary.
Period: [PERIOD] Comparison: [budget / forecast / prior year]
Audience: [BOARD / CEO / DEPT HEAD]
Materiality threshold: [$ AMOUNT OR %]
Variance data:
"""
[PASTE: line item, actual, budget, variance $, variance %]
"""
Known drivers / one-time items:
"""
[PASTE OR "none provided"]
"""
Produce, following templates/executive-narrative-template.md:
1. HEADLINE: one sentence on the overall result vs plan.
2. MATERIAL VARIANCES: for each above threshold, a sentence naming the driver
in business terms and whether it is favorable/unfavorable.
3. ONE-TIME VS RECURRING: separate them clearly.
4. SO WHAT: implication for the forecast or recommended action.
Rules:
- Exclude variances below the materiality threshold.
- Explain drivers in business language, not "due to timing".
- Do not invent drivers; if unknown, say "driver not yet identified".
Quality checklist
- Only material variances are discussed.
- Each variance names a real business driver, not "timing".
- One-time items are separated from recurring trends.
- Favorable vs unfavorable is explicit for each item.
- The narrative ends with an implication or action.
- No driver was invented; unknowns are labeled.
Common mistakes
- Restating the variance numbers as prose without explaining why.
- Treating a one-time item as a trend (or vice versa).
- Burying the one variance that actually matters among trivial ones.
Example output
Headline: Operating income beat plan by $1.2M, driven mostly by delayed hiring.
Material variances:
- Payroll favorable $0.9M: 6 planned hires slipped to next quarter (recurring risk).
- Cloud costs unfavorable $0.3M: one-time migration spend.
One-time vs recurring: migration spend is one-time; hiring delay will reverse.
So what: the beat is largely timing. Hold forecast flat and expect payroll
to normalize next quarter.
Related skills
- Spreadsheet Anomaly Review — to catch data issues before writing the narrative.
- Board Deck Insight Extractor — to pull variances out of an existing deck.
- Decision Memo Builder — when a variance demands a decision.
Attribution
This skill was created by Vectory and is licensed under CC BY 4.0.
Source: https://vectory.io/skills/finance-variance-narrative
Attribution: "Finance Variance Narrative Generator" by Vectory.