A spreadsheet full of numbers does not explain itself. You have rows of data, a question to answer, and no clear idea where to begin.
This guide gives you ChatGPT prompts for data analysis that help you prepare your data, explore it, and report what you found. You will get prompts for each step, plus the mistakes to avoid, like sharing private data or trusting the math without checking. AI can guide your thinking, but you stay in charge of the numbers.
Use the cheat sheet below as a quick reference. Then read on to see when to use each prompt and how it looks filled in.
How to Use These Prompts
Words in [brackets] are blanks. Replace each one with your own details before you send the prompt.
Before: Describe this data: [sample]
After: Describe this data: Date, Product, Units, Price. Jan 3, Mug, 12, 8.50. Jan 4, Tote, 5, 15.00.
Paste a small sample of made-up or anonymized rows, not your whole private dataset. These prompts work the same way in ChatGPT, Claude, and Gemini. The AI can make math errors, so recheck every number in a spreadsheet or calculator.
Prepare Your Data
Describe this data: [sample]
Start by checking that the AI understands your data. Paste a few rows and ask it to describe the columns and types. If the description is wrong, fix your sample or add notes.
Example: Describe this data: Date, Product, Units, Price. Jan 3, Mug, 12, 8.50. Jan 4, Tote, 5, 15.00.
Spot problems in this data
Messy data leads to messy results. This prompt flags blanks, duplicates, and odd values in the sample you provide. Review each flag, since the AI may miss problems that are not in the sample.
Example: Spot problems in this data: Jan 3, Mug, 12, 8.50. Jan 3, Mug, 12, 8.50. Jan 5, Tote, , 15.00. Jan 6, Mug, -4, 8.50.
Suggest how to clean [column]
Dates, names, and numbers often come in mixed formats. Describe the column and the mess, and ask for cleaning steps. Test any steps on a copy of your data first.
Example: Suggest how to clean the Date column, which has entries like Jan 3, 1/3/2025, and 3 January.
Columns: [list], goal: [x]
The AI cannot see what your columns mean. Spell them out and state your goal. A clear goal keeps the analysis focused on your real question.
Example: Columns: Date, Product, Units, Price. Goal: find out which product sells best each month.
Use made-up sample rows
You can test a method without exposing private data. Ask the AI to create fake rows that match your structure. Run your analysis on the fake rows first.
Example: Use made-up sample rows that match this structure: Date, Product, Units, Price. Create 10 rows for a small online shop.
Explore the Data
What questions can I answer?
If you have data but no question, this prompt suggests what you could look into. Give your columns and the context. Choose questions that matter to your goal.
Example: What questions can I answer with columns for Date, Product, Units, and Price for a small online shop?
Find trends in [data]
Paste numbers and ask the AI to describe patterns. It can point out rises, drops, and repeating shapes. Check each trend with a chart or a calculation, since a small sample can mislead.
Example: Find trends in this data: Jan 120 units, Feb 135, Mar 150, Apr 140, May 160.
Compare [group A] and [group B]
Comparisons help you see differences. Define the groups and the measure you care about. Ask the AI to note limits, such as groups of very different sizes.
Example: Compare weekday sales and weekend sales for a small online shop. Weekday average is 45 units, weekend average is 70.
Find outliers in [column]
Outliers are values far from the rest. They can be errors or real events. The AI can point them out, but you decide whether they are mistakes by checking the source.
Example: Find outliers in the Units column: 12, 14, 11, 13, 90, 12, 15.
Which chart fits this data?
The right chart makes a point clear. Describe your data and your question, and ask which chart type fits. Then build it yourself in a spreadsheet.
Example: Which chart fits this data? I want to show monthly sales for three products over six months.
Report the Findings
Explain these results simply
Numbers need words. This prompt turns results into plain-language sentences. Compare the sentences with your figures to make sure nothing was changed.
Example: Explain these results simply: Mug sales rose from 120 to 160 units from January to May, while tote sales stayed near 50.
Write 3 key takeaways
Busy readers want the main points. Ask for three takeaways based only on the data you gave. Cut any that go beyond what the data shows.
Example: Write 3 key takeaways from this monthly sales summary, using only the numbers I provided.
Draft a summary for [audience]
A manager, a client, and a teammate may want different levels of detail. Name the audience and ask for a short summary. Edit it so it fits your workplace.
Example: Draft a summary for the shop owner, who wants a short update with no jargon, about monthly sales trends.
List caveats for this analysis
Every analysis has limits, like a small sample or missing months. This prompt lists them so readers are not misled. Add the real limits you know about.
Example: List caveats for this analysis, which uses five months of sales for one small online shop.
Next steps from [result]
Findings are most useful when they lead to action. Ask for ideas worth testing, not guaranteed fixes. Choose steps you can actually carry out.
Example: Next steps from the result that mug sales are growing faster than tote sales.
Common Mistakes
Uploading sensitive data
Do not paste names, ID numbers, contact details, or private financial figures into an AI tool. Remove or replace them first. Quick fix: use anonymized IDs like Customer 1.
Trusting the math
AI can make arithmetic errors, especially with longer lists. Recheck totals and averages in a spreadsheet or calculator. Use the AI for ideas and explanations, not final numbers.
Ignoring data quality
Bad input leads to bad output. Clean errors, duplicates, and blanks before you analyze. A quick scan of your data can save hours later.
Mistaking trends for causes
A pattern does not prove why something happened. Sales may rise for many reasons. Use findings as clues, and look for more evidence before you decide.
Skipping a sample check
Before you trust a method on your full data, test it on a few rows where you know the answer. If the results do not match, fix the method first.
Pro Tip
Test every method on a small sample with known answers before you trust it on your full data. If the AI’s result does not match your own, find out why first.
Example: Here are five rows and the totals I calculated by hand. Show your method, and tell me if your result matches mine.
FAQ
Can ChatGPT analyze spreadsheet data?
It can help you plan an analysis, explain results, and suggest methods, especially with small samples you paste in. It can make mistakes, so recheck numbers yourself and avoid sharing private data.
Is it safe to share my data with AI?
Be careful. Remove names, IDs, and private figures before pasting anything, and check the privacy settings of the tool you use. Anonymized or made-up samples are safer.
What should I do before I analyze data?
Clean it. Check for blanks, duplicates, and odd values, and make sure each column has a clear meaning. Then state your question so the analysis has a clear goal.
Final Thoughts
AI can help you organize your thinking about data, but it does not replace careful checking. Start with a small sample, test your method, and recheck every number. Protect private information at every step.
If you want the full prompt pack for working with data, get it here: Get the prompt pack here