ChatGPT Prompts for Data Analysis

Did B actually win, or are you fooling yourself? Give it the numbers — get whether it's real, how confident, and what to do next. No false positives.

Building something and need fake-but-realistic data? Describe the shape — get a clean sample dataset with believable values and the edge cases to test.

Two lists that should match but don't. Paste them — get what's in one but not the other, the duplicates, and the near-matches you'd miss by eye.

Paste your numbers over time — get a straight-talk projection of where they're heading, what could throw it off, and how much to trust it.

Say what you want in plain English, get clean SQL back — plus the gotchas (nulls, dupes, slow joins) you'd only find in production at 2am.

Tracking everything means tracking nothing. Tell it your business — get the handful of numbers that actually matter and how to lay them out.

Hundreds of open-text responses you'll never read? Paste them — get the real themes, the quotes that capture them, and what to actually do about it.