Budget vs. Actual Analysis

I'm uploading our P&L with actuals and budget for ${month}. Analyze every line item and produce:

1. **VARIANCE TABLE**
Format as:
| Line Item | Actual | Budget | $ Variance | % Variance | Favorable? | Explanation |

Only include the "Explanation" column for variances over ${threshold}% or $${dollarThreshold}.

2. **MATERIALITY FILTER**
Separate the variances into:
- Material (over ${threshold}% AND over $${dollarThreshold}) — these need explanation
- Immaterial — note these exist but don't waste the board's time

3. **NARRATIVE SUMMARY** (3-5 paragraphs)
For each material variance:
- What happened (the number)
- Why it likely happened (timing, volume, price, one-time vs recurring)
- Whether it's a concern or expected
- What action is recommended (if any)

4. **ONE-LINE BOARD SUMMARY**
A single sentence a CEO would text to their board chair. Example: "February revenue beat plan by 8% on enterprise expansion; marketing overspend by 15% is timing-related and will normalize by Q2."

Tone: direct, specific, no hedging. Use actual numbers. Frame variances in business terms, not accounting terms.

Here's a variance that needs explanation for the board:

Line item: ${lineItem}
Actual: $${actual}
Budget: $${budget}
Variance: $${variance} (${variancePct}%)

Context I know:
${context}

Write a 2-3 sentence explanation that:
1. States the variance in plain terms
2. Explains the driver(s)
3. Categorizes it: timing shift, volume-driven, price-driven, one-time, or structural change
4. Recommends whether to adjust the forecast or monitor

Write it in the voice of a CFO presenting to a board — concise, confident, specific.

I'm uploading our variance report. For each line item with a variance over ${threshold}%, classify it into one of these categories:

1. **Timing** — happened earlier/later than planned, will self-correct
2. **Volume** — more/fewer units than planned
3. **Price/Rate** — different unit cost than planned
4. **One-Time** — non-recurring, exclude from run-rate
5. **Structural** — permanent change, need to reforecast

For each, write one sentence explaining the classification.

Then summarize: "Of the $X total unfavorable variance, $Y is timing-related and will self-correct, $Z is structural and requires a forecast update, and $W is one-time."

This framing helps the board understand whether the variances are noise or signal.

I'm uploading our YTD P&L (${months} months of actuals vs budget). Produce:

1. **YTD VARIANCE SUMMARY**
Same table format as monthly, but cumulative. This smooths out timing variances and reveals structural issues.

2. **TREND ANALYSIS**
For the top 5 largest YTD variances, show the monthly progression:
- Month 1: +$X
- Month 2: +$Y (cumulative: +$Z)
- Month 3: ...

Identify whether the variance is:
- Growing (getting worse each month — structural problem)
- Stable (same delta each month — permanent but not worsening)
- Converging (self-correcting — likely timing)

3. **FORECAST IMPACT**
Based on YTD trends, estimate the full-year variance for each material line item. "At current run-rate, Engineering will finish the year $180K over budget. If the contractor engagement ends in April as planned, the full-year overage reduces to $90K."

4. **REFORECAST RECOMMENDATIONS**
List line items where the budget should be updated, with suggested new numbers and justification.

Take this variance data and format it for a board deck slide. I need:

1. **Headline** — One sentence summarizing the month. Example: "Revenue ahead of plan; opex timing shift will normalize Q2."

2. **Key Metrics Bar** — Show these 4 metrics in a horizontal row:
 - Revenue: $X (vs plan: $Y, +Z%)
 - Gross Margin: X% (vs plan: Y%)
 - EBITDA: $X (vs plan: $Y)
 - Cash: $X (runway: Y months)

3. **Variance Waterfall** — Describe a bridge from budgeted EBITDA to actual EBITDA:
 Budgeted EBITDA → Revenue beat → COGS variance → ... → Actual EBITDA
 (List each component with + or - and the driver)

4. **Traffic Light Summary** — For each department or major line item:
 🟢 On/ahead of plan
 🟡 Within 5% of plan, monitoring
 🔴 Off plan, action required

Keep the total content to what fits on one slide. Ruthlessly prioritize.

Build a Python script that reads two CSV files — actuals and budget — and generates a variance analysis HTML report.

INPUT FILES:
- actuals.csv: columns = category, subcategory, amount
- budget.csv: same structure

THE SCRIPT SHOULD:
1. Merge actuals and budget on category + subcategory
2. Calculate $ variance and % variance
3. Flag material variances (over ${threshold}% AND over $${dollarThreshold})
4. Generate an HTML report with:
 - Summary metrics at top (total revenue variance, total expense variance, EBITDA variance)
 - Detailed variance table with conditional formatting (green/red)
 - A section for "Material Variances Requiring Explanation" — list each material variance with a placeholder for the narrative (I'll fill these in)
5. Sort the material variances by absolute dollar impact, largest first

Save as variance-report-YYYY-MM.html.