26 Copilot Prompts for Excel (2026)

Before you paste any of these, format your data as an Excel Table. Select the range and press Ctrl+T (or Insert → Table) so every column has a clear header and the data is one continuous block. Microsoft 365 Copilot in Excel works on structured tabular data, not scattered cells — a real Table lets it reference your columns by name, reason over the whole dataset, and insert formulas, PivotTables and charts cleanly. Wherever a prompt shows a placeholder like [Revenue] or a range like A2:D200, swap in the exact column headers from your own Table.

The 26 prompts below are grouped into five jobs: writing formulas, surfacing insights and trends, building PivotTables and charts, cleaning and formatting messy data, and running deep Analyst agent analysis. Each is copy-paste ready — refine it with the GCSE framework (Goal, Context, Source, Expectations) covered in how to prompt Copilot for work. For the full cross-app collection, start with the 35 best Microsoft Copilot prompts, and keep the Copilot prompt cheat sheet open while you work.

Formulas & formula columns

Copilot can write, explain and fix formulas, and add whole calculated columns to your Table. Describe the calculation in plain English and name the columns it should use — Copilot returns the formula and offers to insert it as a new column.

1. Profit margin percentage column

Add a column called Profit Margin that calculates (Revenue − Cost) / Revenue for each row, using the [Revenue] and [Cost] columns, and format the result as a percentage with one decimal place.

Why it works: it names the exact columns, gives the formula in words, and specifies the output format, so Copilot inserts a ready-to-use calculated column.

2. Look up a value with XLOOKUP

Add a column that looks up the [Region] for each [Sales Rep] by matching against the reps listed on the Territories sheet, using XLOOKUP so it returns "Unassigned" when there is no match.

Best for: pulling data across sheets without hand-writing a fragile VLOOKUP.

3. Categorize rows with a nested IF

Add a column called Deal Size that labels each row based on the [Amount] column: "Large" if Amount is 50000 or more, "Medium" if between 10000 and 50000, and "Small" if under 10000. Use an IFS formula.

Why it works: spelling out the thresholds and the desired labels lets Copilot build the branching logic exactly, and asking for IFS keeps it readable.

4. Days between two dates

Add a column called Days to Close that returns the number of days between the [Created Date] and [Closed Date] columns as a whole number, and leave it blank if [Closed Date] is empty.

Best for: cycle-time and turnaround metrics from a pair of date columns.

5. Explain and fix a broken formula

Explain in plain English what the formula in cell F2 does, tell me why it is returning a #REF! error, and give me a corrected version that works across the whole [Total] column.

Why it works: Copilot can read, describe and repair existing formulas — asking for the explanation plus the fix means you learn the cause, not just the patch.

6. Conditional total with SUMIFS

Write a formula that sums the [Revenue] column only for rows where [Region] equals "West" and [Status] equals "Closed Won", and explain each argument so I can adapt it to other regions.

Best for: a single conditional number without building a full PivotTable.

Analysis, insights & trends

Point Copilot at the Table and ask it to interpret the numbers. It can summarize, rank, compare, detect trends over a date column and flag outliers — then explain what it found in words. Reference the exact columns so it analyzes the right fields.

7. Summarize the key insights

Analyze this table and give me the four most important insights about [Revenue] across [Region] and [Product Category]. Write each as a one-sentence bullet with the specific numbers that support it.

Why it works: capping it at four bullets and demanding supporting numbers forces a concise, evidence-backed read instead of a vague summary.

8. Identify top and bottom performers

Using the [Sales Rep] and [Revenue] columns, tell me the top five and bottom five reps by total revenue, and note how far each group is from the overall average.

Best for: a fast leaderboard read from a raw transaction list.

9. Spot trends over time

Look at the [Order Date] and [Revenue] columns and describe the month-over-month trend for the last 12 months. Call out which months grew, which declined, and whether there is any seasonality.

Why it works: naming the date and value columns and the window tells Copilot exactly how to slice time so the trend read is grounded in your data.

10. Flag outliers and anomalies

Scan the [Order Amount] column and list any rows that look like outliers or data-entry errors — unusually high or low values, or amounts that don't fit the pattern for their [Product Category]. Show the row and why it stands out.

Best for: catching bad records before they skew a report.

11. Compare two segments

Compare [New] versus [Returning] customers in the [Customer Type] column across average [Order Value], total [Revenue], and order count. Summarize the biggest differences in three bullets.

Why it works: a defined split and a defined set of metrics keeps the comparison tight and repeatable.

12. Ask a plain-English question

Which [Product Category] had the highest total [Revenue] in Q4, and what percentage of overall Q4 revenue did it represent? Answer with the number and a short explanation.

Best for: answering an ad-hoc business question without writing a formula yourself.

PivotTables & charts

Copilot can generate PivotTables and charts directly from your Table and drop them onto the sheet. Tell it the field to summarize, how to group it, and how you want it visualized — then refine the result with follow-up prompts.

13. PivotTable of revenue by region and month

Create a PivotTable that shows total [Revenue] with [Region] as rows and the month of [Order Date] as columns, sorted so the highest-revenue region is at the top.

Why it works: it specifies the value, the rows, the columns and the sort, so Copilot builds the exact matrix rather than guessing the layout.

14. PivotTable of average order value by segment

Build a PivotTable showing the average [Order Value] and the count of orders for each [Customer Segment], and add a grand total row at the bottom.

Best for: a compact summary table you can drop into a report.

15. Column chart of sales by category

Create a column chart of total [Revenue] by [Product Category], sorted from highest to lowest, with data labels showing the value on each bar.

Why it works: naming the chart type, the fields, the sort order and the labels gives Copilot everything it needs to insert a finished visual.

16. Line chart of the monthly trend

Make a line chart of total [Revenue] by month using the [Order Date] column, and add a linear trendline so I can see the overall direction.

Best for: showing momentum over time in a single glance.

17. PivotTable with percentage of total

Create a PivotTable of total [Revenue] by [Sales Channel], and add a second column that shows each channel as a percentage of the grand total, sorted largest first.

Why it works: asking for the percentage-of-total column turns a plain sum into a share-of-mix view without extra formulas.

Clean, format & organize

Messy data is the top reason Copilot gives weak answers. Use it to standardize text, split and reformat columns, highlight problems, and sort or filter to a working view — all before you analyze. Clean the Table first, then ask your questions.

18. Standardize inconsistent text values

The [Country] column has inconsistent values like "USA", "U.S.A.", "United States" and "us". Standardize them all to a single consistent label per country and put the cleaned values in a new column called Country Clean.

Best for: merging variant spellings so grouping and filtering actually work.

19. Split full name into first and last

Split the [Full Name] column into two new columns, First Name and Last Name, handling middle names by keeping everything after the first word in Last Name.

Why it works: stating the rule for middle names removes the ambiguity that usually breaks a naive split.

20. Highlight duplicates and blanks

Highlight any rows where the [Email] value is duplicated, and separately highlight any rows with a blank [Email] or blank [Order ID], so I can review data-quality problems.

Best for: a quick visual audit before you trust the dataset.

21. Reformat dates to one format

The [Signup Date] column mixes formats like "3/5/26", "March 5 2026" and "2026-03-05". Convert them all to real Excel dates in a new column formatted as YYYY-MM-DD, and flag any values that can't be parsed.

Why it works: asking Copilot to flag unparseable values means bad dates surface instead of silently becoming wrong ones.

22. Sort and filter to a working view

Filter the table to show only rows where [Status] is "Open" and [Priority] is "High", then sort by [Due Date] from soonest to latest.

Best for: carving a large Table down to the rows you need to act on.

The Analyst agent (deep data analysis)

The Analyst agent goes beyond formulas: it reasons step by step and can write and run Python on your data for forecasting, segmentation, correlation and statistical work, showing its calculations as it goes. Your Copilot license includes roughly 25 combined Analyst and Researcher advanced queries per month, so save these for the harder questions.

23. Forecast next quarter

Using the Analyst agent, forecast total [Revenue] for the next three months based on the monthly history in the [Order Date] and [Revenue] columns. Account for any seasonality, give a confidence range, and show the method you used.

Why it works: invoking the Analyst and asking it to show its method turns a guess into a reproducible, explained forecast you can defend.

24. Segment customers into groups

Use the Analyst agent to cluster customers into distinct segments based on total [Spend], order [Frequency], and [Recency] in days. Name each segment, describe what defines it, and tell me how many customers fall into each.

Best for: RFM-style segmentation without exporting to a separate tool.

25. Run a correlation analysis

Use the Analyst agent to measure the correlation between [Marketing Spend] and [Revenue] across the rows in this table. Report the correlation strength, whether it looks statistically meaningful, and caution me about any confounders.

Why it works: asking for statistical meaning and confounders keeps the agent honest instead of implying a spurious cause.

26. Build a driver analysis

With the Analyst agent, analyze which columns most strongly explain differences in [Churn] across customers. Rank the top drivers, quantify each one's contribution, and summarize what the data suggests we should act on.

Best for: understanding what actually moves a KPI before you build a model around it.