Copilot Prompts for Excel That Clean Up Workbook Work

Tested on: Microsoft 365 Copilot with Excel for Microsoft 365 in a work or school account. Copilot buttons, table behavior, and available actions can vary by license, tenant policy, app version, and rollout ring.

Good Excel Copilot prompts do more than ask for a formula. They describe the workbook area, the business question, the output you want, and the checks you need before accepting the result. That matters because Excel work is often used for decisions, reports, and follow-up actions where a polished answer can still be wrong.

Use the prompts below as reusable patterns for normal spreadsheet work: analyzing a table, creating formulas, finding outliers, explaining trends, cleaning columns, and turning a workbook into an action list. Replace the bracketed details with your real table names, column names, and business context.

Prepare the workbook before prompting

Copilot works best when the data range is easy to understand. Give your table clear headers, remove blank header rows, and avoid mixing several unrelated lists on the same worksheet. If the workbook has multiple tables, select the one you want Copilot to use before asking for analysis.

A reliable starting prompt is: “Use the selected table. Identify the main trends in [metric], explain the likely drivers, and list any rows that need review. Do not change the workbook yet.” This gives Copilot context and keeps the first response in review mode.

Prompt for a clean first read

  • Summarize this table in five bullets for a manager who needs the business result, not the spreadsheet details.
  • Identify the top three changes in [metric] by [month/region/team] and explain what might need follow-up.
  • List missing, duplicate, or unusual values in the selected table before suggesting formulas or charts.
  • Tell me which columns are needed to answer this question: [business question].

Prompts for formulas and calculations

Formula prompts should name the calculation and the columns involved. Avoid asking for “a formula for this” unless the selected cell and source columns are obvious. A better prompt states the goal, where the source values live, and whether you want the formula inserted or only explained.

Excel Copilot prompt workflow for workbook analysis
Use a clear Excel prompt to select the table, ask the question, choose an output, and review changes before applying them.

Use: “Create a formula that calculates [result] from [column A] and [column B]. Explain the formula first, then suggest where it should go. Do not insert it until I approve.” This gives you a review step before Copilot modifies the workbook.

Useful formula prompt patterns

  1. Create a formula to flag rows where [date column] is older than [number] days and [status column] is not complete.
  2. Suggest a formula that categorizes [amount column] into low, medium, and high bands using these thresholds: [thresholds].
  3. Write a formula that extracts the domain from [email column] and explain how it handles blank cells.
  4. Build a summary formula for total [sales/cost/hours] by [region/team/month], then show the result as a small table.

For users who do not see the Copilot button in the workbook, the first troubleshooting step is to that assistant setup path for the first check.

If the same symptom points to account access rather than the current screen, the source-context checks gives you a useful comparison point before you change more settings.

Prompts for analysis, charts, and PivotTables

When you want analysis, tell Copilot whether you need an explanation, a chart, a PivotTable, or a new calculated column. Those are different outputs. If you ask only for “insights,” Copilot may produce a useful summary, but it may not create the artifact you need for the next meeting.

Ask for the output explicitly

  • Find the largest month-over-month changes in this table and suggest a chart type that would make them easy to explain.
  • Create a PivotTable-style summary of [metric] by [category] and [month]. Show the proposed layout before inserting it.
  • Highlight rows that look like outliers and explain which column made each row unusual.
  • Compare [current period] to [previous period] and write a short explanation I can paste into a status update.

If the analysis will be shared, ask for a short plain-English explanation after the chart or PivotTable suggestion. That forces Copilot to connect the workbook output back to the business question.

Prompts for cleaning and standardizing data

Data cleanup prompts should be conservative. Ask Copilot to identify the cleanup plan before it changes values. This is especially important for customer names, product codes, dates, currency, and imported CSV data where small differences may be meaningful.

Use: “Review this table for inconsistent values in [column]. Group likely matches, explain the suggested standard value, and wait for approval before changing anything.” That prompt works better than “clean this column” because it protects valid exceptions.

Guardrails for cleanup prompts

  • Ask Copilot to list affected rows before applying changes.
  • Tell Copilot not to overwrite the original column until you approve the cleanup.
  • Request a helper column when you need to audit the result.
  • Confirm date formats, decimal separators, and currency symbols manually.

Review before applying workbook changes

For serious workbooks, end every prompt with a review instruction. Ask Copilot to explain what it will change, which cells or columns are affected, and how you can verify the result. If the output includes formulas, test the first few rows manually before filling the entire column.

A good final review prompt is: “Before applying changes, list the workbook edits, affected range, formulas added, and checks I should perform. Flag any assumption you made from the data.” This turns Copilot from a quick answer tool into a controlled workbook assistant.

Reader questions

What is the best Copilot prompt for Excel formulas?

The best prompt names the source columns, the calculation, and the review step. For example, ask Copilot to create a formula for a specific result, explain it, and wait before inserting it into the worksheet.

Can Copilot analyze any Excel workbook?

Copilot is most reliable when the data is structured as a clear table with headers. Very messy sheets, merged cells, hidden assumptions, and mixed data blocks make it harder for Copilot to infer the right context.

Should I trust Copilot-generated Excel formulas?

Treat formulas as suggestions. Check the referenced columns, test a few rows, compare totals against a known result, and keep a copy of the original data when the workbook supports reporting or financial decisions.

Before you move on

Strong Excel Copilot prompts are specific because workbook mistakes are easy to hide inside a neat table. Name the table, the question, the output, and the review step. Then verify the formula, summary, chart, or cleanup before using it in a decision.