Clean Up Messy Data by Just Asking

You don't need a single spreadsheet formula to fix messy data anymore. You need a clear description of what "clean" looks like, and a chat window.

Start by describing the mess

Paste a sample of your data — not the whole file, just enough rows to show the problem — and describe what's wrong in plain words: "Some dates are written three different ways, some names have extra spaces, and a few rows are duplicated." The clearer your description, the better the fix.

Ask for the fix, not just the answer

Instead of "clean this up," try: "Standardize every date to YYYY-MM-DD, trim extra spaces from names, and flag any row that looks like a duplicate so I can review it before deleting anything." Naming the exact rules gets you a consistent result instead of a guess.

Let it explain its logic first

For anything you'll reuse — a template, a recurring report — ask the AI to explain its cleanup rules before applying them: "Before you change anything, list the rules you're going to apply." This catches misunderstandings early, when they're a one-line fix instead of a re-do.

Keep a human in the loop for deletions

AI is excellent at spotting likely duplicates or errors, but let it flag rather than delete. Ask it to add a column marking rows it's unsure about, so you make the final call on anything that isn't obviously wrong.

A quick example

Say a row reads: " Jon Smith , 3/4/25, jon.smith@gmail .com". Ask: "Clean this contact: trim whitespace, fix the email if there's a stray space, and convert the date assuming US format." You'll get back something like "Jon Smith, 2025-03-04, jon.smith@gmail.com" — plus a note if anything was ambiguous, like whether 3/4 means March 4th or April 3rd.

A workflow that scales

Once you've got a set of cleanup rules that work, save the exact prompt. Next time you get a messy export, paste the new data and the same prompt — you've just turned a one-off fix into a repeatable five-minute routine instead of an afternoon of formulas.