AI Monthly Business Review — Original SKILL.md

Prompt

---
name: monthly-business-review
description: Turn a month of numbers and notes into an honest business review — the month in one line, headline figures, what actually moved, wins, misses, a clearly-labelled hypothesis for the biggest miss, three ranked priorities for next month, the one number to watch, and what data was missing. Use this when the user pastes monthly figures or says review my month, how did last month go, or write my monthly business review.
---

# Monthly Business Review

When this skill runs, turn the numbers and notes the user pastes in into an honest, useful review they can read in five minutes and act on.

## Process

1. Identify the month being reviewed and the type of business from the material provided.
2. Restate the headline numbers exactly as the user gave them. Do not round in a way that flatters or alarms.
3. Compare to the prior period or to a target only if the user provided those figures. If they did not, write "No comparison provided" — do not invent a baseline.
4. Identify what actually moved. Distinguish a meaningful change from normal month-to-month noise, and say which is which.
5. Separate wins (worked, worth repeating) from misses (underperformed, worth fixing).
6. For the single biggest miss, give the most likely explanation — labelled clearly as a hypothesis, not a fact — and state exactly what data would confirm or disprove it.
7. Recommend three priorities for next month, ranked, each tied to something specific in this month's data.
8. Name the one number to watch next month.
9. List what was missing from the input that would have made the review sharper.

## Rules

- Use only the numbers and notes the user provides. Never invent revenue, traffic, leads, conversion rates, churn, headcount, ad spend, or any other metric.
- You cannot connect to Google Analytics, Stripe, QuickBooks, a CRM, or any dashboard. You have no data unless the user pastes it in. Never imply otherwise.
- You **may** do arithmetic on the numbers provided — differences, percentages, ratios, totals. Show the calculation in one short line so the user can check it. Never estimate a number that was not provided.
- **Always label an explanation of "why" as a hypothesis.** One month of data does not establish a cause.
- Do not soften a bad month; an honest review is the entire point. Equally, do not manufacture alarm about ordinary variation.
- If the user provides too little to review properly, say so plainly and list exactly what to gather before next month.

## Output format

### The Month in One Line
The honest headline — including when a flat top-line number is hiding something.

### Headline Numbers
As provided. Include a comparison only if a baseline was given.

### What Moved
Meaningful changes vs. normal noise.

### Wins — Worth Repeating

### Misses — Worth Fixing

### Most Likely Explanation (Hypothesis)
For the biggest miss. Include what would confirm or disprove it.

### Next Month's Top 3 Priorities
Ranked. Each tied to something above.

### The One Number To Watch

### What Was Missing
Data that would have made this review sharper.

If the user has not pasted any numbers, ask what they track before writing anything.