Purpose
Convert a table of actuals versus budget into clear, decision-grade management commentary — explaining the real drivers behind each material variance, distinguishing timing from permanent effects, and framing the forward risks and opportunities a leadership team needs to act on. Eliminates the all-too-common "spending was higher than planned" filler and produces commentary that survives scrutiny in a finance review.
When to Use
Invoke this skill when the user asks to:
- Write management commentary for a monthly or quarterly variance report
- Explain why actuals came in over or under budget (or forecast)
- Draft the narrative section of a board pack or finance review deck
- Turn a P&L variance table into a written CFO-ready summary
- Frame budget overruns or underspends in terms of risks and opportunities
- Summarize variance drivers for a business partner conversation with a department head
Inputs Required
Before starting, collect:
- Reporting period: The month, quarter, or YTD period being reported
- Variance data: Line-item actuals vs. budget (and prior forecast if available), in absolute and/or % terms
- Materiality threshold: The $ or % level above which a variance must be explained (e.g., $50K or 5%)
- Comparison basis: Actuals vs. budget, vs. forecast, vs. prior year — or a combination
- Known drivers: Any context the analyst already knows (e.g., delayed hire, one-off legal fee, FX swing, deal slippage)
- Audience: CFO, board, department head, or executive committee — sets tone and depth
- Reporting level: Whole P&L, a single cost center, a function, or a business unit
If materiality threshold is missing, default to the greater of 5% or $50K. If audience is unspecified, default to CFO-level commentary.
Instructions
Step 1: Filter to Material Variances Only
Apply the materiality threshold and discard the noise.
- List only line items where the variance exceeds the threshold in $ or %
- A small % on a large base, or a large % on a small base, can both be material — flag both
- Suppress immaterial items entirely; commentary that explains a $2K variance erodes credibility
- Rank surviving variances by absolute $ impact, largest first — this becomes the commentary order
Step 2: Classify Each Variance by Type and Direction
For every material variance, tag two attributes:
- Direction: Favorable (F) or Unfavorable (U) — note that under-spend on a cost line is F, but under-delivery on revenue is U
- Nature:
- Timing — the spend or revenue will still occur, just in a different period (reverses out over the year)
- Permanent — a genuine over/under that will not reverse (run-rate change, lost deal, structural cost shift)
- One-off — a discrete non-recurring item (settlement, refund, true-up) Misclassifying timing as permanent is the most common variance-commentary error and leads to wrong forecast conclusions — be explicit.
Step 3: Isolate the Driver, Not the Symptom
For each variance, push past "we were over budget" to the causal driver. Use the price/volume/mix and rate/efficiency decomposition where applicable:
- Revenue variances: split into volume (units/deals) vs. price/rate vs. mix
- Personnel variances: split into headcount (FTE count vs. plan) vs. timing of hires vs. comp/rate
- Cost variances: split into volume of activity vs. unit cost vs. one-off charges
- State the driver in one sentence with the magnitude attributed: e.g., "Marketing was $120K (18%) unfavorable, driven almost entirely by the pull-forward of the Q3 campaign into June ($95K, timing) with the balance from higher agency rates ($25K, permanent run-rate increase)."
Step 4: Frame Forward-Looking Risks and Opportunities
Translate each variance into its implication for the rest of the year:
- Risk: an unfavorable permanent variance or a timing item that signals a deteriorating run-rate — quantify the full-year exposure if it continues
- Opportunity: a favorable variance that could be locked in, or an underspend that creates reinvestment headroom — quantify it
- Where the variance is timing, state explicitly when it reverses so leadership doesn't over-react
- Each risk/opportunity should carry a $ magnitude and a directional full-year impact
Step 5: Recommend Action Where Warranted
For variances that demand a decision, append a recommended action: re-forecast the line, reallocate budget, escalate to the cost-center owner, or accept and monitor. Not every variance needs an action — flag only those that do, and name the owner where known.
Step 6: Write the Commentary
Assemble into the output structure below. Lead with the headline result, then walk the material variances in $ order, then close with the forward view. Use plain finance English — quantify every claim, attribute every driver, and never explain a variance with a tautology ("costs were high because we spent more").
Output Format
- Structure: Headline summary paragraph → material variance walk (one tight paragraph or bullet per item, largest first) → consolidated Risks & Opportunities section → recommended actions
- Length: 500–900 words depending on the number of material variances
- Format: Prose paragraphs for narrative; a compact Risks & Opportunities table or labeled bullet list; optional one-line headline metrics block at the top
- Sections: Executive Headline → Variance Drivers (by line item) → Risks & Opportunities → Recommended Actions
- Tone: Senior FP&A — precise, quantified, plain-spoken, no hedging and no jargon padding; every sentence either states a number, a driver, or an implication
Quality Checklist
Before delivering output, verify:
- [ ] Every variance discussed exceeds the materiality threshold — no immaterial items included
- [ ] Each variance is labeled Favorable/Unfavorable AND Timing/Permanent/One-off
- [ ] Every driver statement carries a $ magnitude attributed to a specific cause — no "spending was higher" filler
- [ ] Timing variances explicitly state when they reverse
- [ ] Each risk and opportunity is quantified with a full-year $ impact
- [ ] Variances are presented in descending order of absolute $ impact
Notes
- Sum the explained variances against the total reported variance — if the material items don't roughly reconcile to the total, an unexplained gap exists and should be flagged rather than hidden.
- The single most valuable move in variance commentary is correctly splitting timing from permanent; it changes whether a number is a problem or a non-event.
- For department-head audiences, lead with the variances they own; for board audiences, lead with the consolidated full-year implication, not the line-item detail.
- When the analyst supplies no known drivers, generate the most probable driver hypotheses but label them as hypotheses requiring confirmation — never present an inferred cause as established fact.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills