Your first task: clean up a messy CSV

Your first task: clean up a messy CSV

Time to make Claude Code do something. We'll take a messy CSV file and ask it to clean it up.

Get the sample file

Download messy_contacts.csv — a typical export with duplicates, inconsistent formatting, and missing fields.

Or use your own: any CSV with messy data will work.

Put the file somewhere simple

Create a folder and put the CSV in it:

mkdir ~/my-first-automation
cp ~/Downloads/messy_contacts.csv ~/my-first-automation/
cd ~/my-first-automation

Run Claude Code

claude

Now type your instruction in plain English:

Read the CSV file in this folder. Clean it up:
- Remove duplicate rows (match by email)
- Standardize phone numbers to (XXX) XXX-XXXX format
- Fill in missing first/last names where the full name exists
- Capitalize names properly
- Sort alphabetically by last name
Save the result as cleaned_contacts.csv

Watch it work

Claude will:

1. Read your CSV
2. Identify the problems
3. Fix them
4. Write a new clean file

This usually takes 10-30 seconds depending on the file size.

Check the result

Open cleaned_contacts.csv — you should see a properly formatted, deduplicated, sorted contact list.

What just happened?

You wrote a few sentences of English, and Claude Code read the file, figured out what was wrong with it, fixed it, and saved the result. You didn't open Excel, you didn't write a formula, you didn't google "how to remove duplicates in CSV."

Key takeaway

Notice how we were specific: "match by email", "format as (XXX) XXX-XXXX", "sort by last name". Vague instructions get vague results. Specific instructions get exactly what you want.

In the next episode, we'll dive deeper into how to write instructions that consistently get great results.