88% of spreadsheets contain errors. Not typos — actual formula mistakes that silently break your numbers. And the average office worker spends 38% of their workday inside Excel, much of it doing things they were never formally trained to do.
Four days ago, OpenAI dropped ChatGPT for Excel — a new add-in that puts GPT-5.4 directly inside your spreadsheets. You describe what you need in plain English, and it builds formulas, cleans data, and updates cells right in your workbook.
But the tool is only as good as what you ask it. Here are 10 prompts I’ve tested that actually produce usable results — not the generic “create a formula” stuff that gives you something you have to rewrite anyway.
Every prompt works with the new Excel add-in, or you can paste them into ChatGPT, Claude, or any other AI assistant and copy the output back.
Formulas That Would Take 20 Minutes to Google
1. The multi-condition sum
Most people know SUMIF. But SUMIFS with three or four conditions? That’s where everyone starts Googling.
Prompt:
Write an Excel SUMIFS formula that totals revenue in column E where the region in column B is “West”, the product category in column C is “Software”, and the date in column A falls between January 1 and March 31, 2026.
This gives you the exact formula with the date range handled properly — something that trips up even experienced Excel users because of how Excel stores dates internally.
2. The lookup that doesn’t break
INDEX-MATCH is better than VLOOKUP in almost every way. But nobody remembers the syntax.
Prompt:
Create an INDEX-MATCH formula that looks up a product name in Sheet2 column A and returns the corresponding price from Sheet2 column C. If no match exists, return “Not Found” instead of an error.
You get a clean formula with IFERROR wrapping. No more #N/A scattered across your reports.
If you work with complex spreadsheets regularly, our Excel Analytics skill has a full library of advanced formula patterns — SUMPRODUCT arrays, dynamic ranges, conditional formatting rules — that go way beyond what a single prompt can cover.
Data Cleaning (The Part Everyone Hates)
3. Fix messy names
You imported a customer list and half the names look like “john DOE”, " Jane Smith “, or “bob jones-williams”. Classic.
Prompt:
Give me an Excel formula to clean a column of names that have inconsistent capitalization, extra spaces, and mixed delimiters. Standardize everything to “First Last” format. The raw names are in column A starting at row 2.
ChatGPT combines PROPER, TRIM, and SUBSTITUTE into one formula. It handles the edge cases that would take you 15 minutes to think through.
4. Validate and extract from messy data
Email addresses, phone numbers, dates in three different formats — imported data is always a mess.
Prompt:
Write formulas to: (1) extract the domain from email addresses in column B, (2) flag any email that doesn’t contain exactly one @ symbol, and (3) standardize phone numbers in column C to the format (XXX) XXX-XXXX, handling entries with or without dashes, spaces, or country codes.
This one usually takes a couple of columns to implement, but ChatGPT maps it all out for you — which formula goes where, what each one does.
For bigger data cleaning jobs — thousands of rows with mixed formats, duplicates, and missing values — our Data Cleaning skill walks you through the full process, including when to use Power Query instead of formulas.
Analysis That Used to Require a Data Analyst
5. Pivot table setup guide
Pivot tables are probably the single most powerful feature in Excel. And most people avoid them because the interface is confusing.
Prompt:
I have sales data with these columns: Date, Product, Region, Salesperson, Revenue, Units Sold. Walk me through setting up a PivotTable that shows monthly revenue broken down by region, with product as a filter. Include how to add a calculated field for average revenue per unit.
ChatGPT gives you click-by-click instructions specific to your data. Not a generic tutorial — actual steps referencing your column names.
6. Spot the trend your boss will ask about
Prompt:
I have 12 months of revenue data in cells A2:B13 (months in A, revenue in B). Write formulas to calculate: month-over-month growth rate, 3-month moving average, and identify which month had the largest single-month decline. Put results in columns C, D, and E with clear headers.
This turns raw numbers into the story your manager actually wants to hear. And it’s the kind of analysis that takes 30 minutes to set up manually if you’re not sure which formula handles moving averages (it’s AVERAGE with OFFSET, and nobody remembers that).
Want to go deeper on turning spreadsheet data into something people actually want to look at? Our Data Storytelling skill covers how to pick the right chart, write the narrative, and present analysis that drives decisions.
Charts and Reports
7. Build the chart your data needs
Picking the right chart type is half the battle. A pie chart for 15 categories? Terrible. A line chart for two data points? Pointless.
Prompt:
I have monthly revenue and expenses for 2025-2026 in columns A through C (24 rows). Write step-by-step instructions to create a combo chart: bars for revenue, a line for expenses, and a secondary axis for profit margin percentage. Use a professional color scheme — no default Excel blue.
The step-by-step instructions work because they’re specific to your layout. And the color scheme tip matters more than you’d think — default Excel charts scream “I spent 30 seconds on this.”
For a systematic approach to choosing chart types based on your data and audience, check out our Chart Selection Guide.
8. Automate a monthly report template
Prompt:
Create an Excel template structure for a monthly sales report with these sections: (1) Executive summary with KPI cards showing revenue, units, and YoY growth, (2) Regional breakdown table, (3) Top 10 products table, (4) Month-over-month trend chart data. Tell me which cells to use for each section and what formulas to include so that changing the month dropdown in cell B1 updates everything automatically.
This is the kind of prompt where ChatGPT saves you the most time. Building a dynamic report template from scratch takes hours. Getting the formula structure from AI and then tweaking the layout? Maybe 30 minutes.
Our free AI for Spreadsheets course walks through building these kinds of automated reports from start to finish — 8 lessons, about 2.5 hours total.
VBA and Automation
9. Format every sheet at once
You’ve got a workbook with 12 tabs — one per month — and they all need consistent formatting. Doing it by hand is mind-numbing.
Prompt:
Write a VBA macro that loops through every worksheet in my workbook and applies these changes: bold the first row (headers), set alternating row colors (white and light gray), auto-fit all columns, freeze the top row, and add a thin border to all cells with data. Skip any sheet named “Summary.”
Copy the output into the VBA editor (Alt+F11 → Insert → Module → paste → run). Done. Twelve sheets formatted in 3 seconds.
Note: The new ChatGPT for Excel add-in doesn’t support VBA yet. Use this prompt in the ChatGPT web app or any AI assistant, then paste the macro into Excel’s VBA editor.
10. Export sheets as PDFs automatically
Prompt:
Create a VBA macro that exports each worksheet as a separate PDF to a folder called “Reports” on my desktop. Name each PDF with the sheet name and today’s date (format: SheetName_2026-03-09.pdf). Skip any hidden sheets.
This one prompt replaces what would be 15 minutes of File → Print → Save As → repeat for every tab. It’s the kind of small automation that compounds — if you run this report monthly, you just saved yourself three hours a year on a single task.
If you want to automate more of your reporting workflow — scheduling, data pulls, distribution — our Spreadsheet Mastery course covers the full pipeline.
ChatGPT for Excel vs. Microsoft Copilot
This is the question everyone asks. Here’s the honest answer.
Microsoft Copilot is built into Excel natively. It can create charts, generate formulas, summarize data, and sort/filter — all without leaving the app. It’s deeply integrated and works well for straightforward tasks inside a single workbook.
ChatGPT for Excel is an add-in, not native. But it’s powered by GPT-5.4, which has stronger general reasoning. It can work across multiple sheets, trace how formulas connect across a model, and explain why something changed — not just do it. It also has direct connections to financial data providers like FactSet and S&P Global, which is a big deal for finance teams.
The practical take: Copilot is better for quick, simple tasks inside your workbook. ChatGPT is better for complex analysis, multi-sheet reasoning, and anything where you need to understand the “why” behind the numbers. Many teams are using both — Copilot for daily work, ChatGPT for deeper analysis.
What ChatGPT for Excel Can’t Do (Yet)
The add-in is still in beta, and some features aren’t supported:
- No VBA or Office Scripts — you still need to use ChatGPT’s web app for macro generation
- No Power Query or Pivot Data Models — complex data transformations still need manual setup
- No data validation rules, named ranges, or slicers — the advanced features that power-users rely on
- Very large workbooks may hit limits — if your file has 100K+ rows, ChatGPT might only process part of it
- US, Canada, and Australia only for now — other regions coming later
These gaps will close. But for now, the most practical workflow is: use the add-in for formula generation and data analysis, use ChatGPT’s web app for everything else (VBA, templates, complex instructions).
Getting Started
If you’re new to using AI with spreadsheets, start with prompts 1, 3, and 5 above — they cover the three tasks that waste the most time for most people: writing formulas, cleaning data, and building pivot tables.
For a structured walkthrough, our free AI for Spreadsheets course covers all of this and more in 8 bite-sized lessons. And if you want a deeper skill library for ongoing Excel work, the Excel Analytics skill is a good one to keep bookmarked.
The best thing about these prompts? They work whether you use the new Excel add-in, ChatGPT in a browser tab, Claude, Gemini, or anything else. The AI doesn’t care. Your spreadsheet doesn’t care. You just stop spending your afternoon debugging nested IFs.