Data Cleaning Plan

I have a messy dataset. Write a cleaning plan.

Ask for:
- The tool I'm using (Excel/Sheets, Python + pandas, R, SQL, other)
- A sample: 5-10 rows and the column names, or a description of the shape
- What I plan to DO with the cleaned data (analysis, chart, load into another system)

Output:

**What's wrong** — bullet list of the specific issues you see. Not "the data is messy," but "column X has mixed date formats" or "duplicates in rows 4, 7, 12."

**Priority-ranked cleaning steps** — numbered list. Each step:
- What to do
- The exact tool command / formula / function to do it
- What to check after to confirm it worked

**What to LEAVE ALONE** — things that look like errors but might not be. Trailing whitespace in some fields, weird outliers that are real, etc.

**What I can't tell without seeing more** — data quality issues you can't judge from a sample alone. Row counts by category, distribution shape, missing patterns.

Rules:
- Never suggest deleting rows/columns without saying what to save first.
- Prefer non-destructive steps (create a cleaned copy) over destructive (overwrite).
- If the data has structural problems (wrong shape entirely), address that before cell-level cleanup.

## How to use it

Paste this prompt, then paste sample rows. Follow the steps in order. Save a "clean" version alongside the raw one so you can rerun.

## Why it works

Cleaning-as-you-go loses. Cleaning against a written plan catches the failure modes you'd otherwise discover mid-analysis. The "leave alone" section prevents the common overshooting where cleaning removes real signal.