Excel Clean Data Skill rolls out in Copilot

Key insights

• A new Clean Data skill is rolling out under Copilot in Excel's Data & Analysis category.
• Users can @mention the Clean Data skill instead of finding the feature through menus.
• It can inspect capitalization, spacing, text-as-number issues, invalid values, duplicates, and outliers.
• The skill is meant to reduce time spent locating tools and writing prompts.
• Microsoft presents it as a Copilot skill for faster, guided data preparation.

Ami Diamond [MVP] recently highlighted a new Clean Data skill in Copilot for Excel. The takeaway is simple: instead of hunting through menus or crafting a prompt from scratch, users can call the skill directly with an @mention and ask Copilot to inspect the data. The post frames this as part of a broader shift toward Copilot skills that make common tasks easier to start.

The update sits under the Data & Analysis category in Copilot in Excel. According to the material, the skill is rolling out, which means readers should treat it as a staged change rather than something that is already universal. The core value is not a new report or charting feature, but a faster path to cleaning the data that those outputs depend on.

What's new

The main change is a new Clean Data skill in Copilot in Excel. The post says that before this change, users had to go to Data → Clean Data, use the yellow Clean Data banner, or write a prompt asking Copilot to clean the data. Now the skill can be called directly with an @mention.

That matters because it changes the entry point. Instead of treating data cleanup as a separate feature to locate, it becomes something a user can ask Copilot to do in the same chat flow used for other tasks. The material also places the skill in the Data & Analysis category, which helps explain where Microsoft has grouped it inside Copilot.

The sources describe the skill as being able to inspect and fix a broad set of common spreadsheet problems. These include inconsistent capitalization and spacing, category-label variations and typos, numbers or dates stored as text, inconsistent visible formats, invalid identifiers, dates, emails, or URLs, invisible and lookalike characters, encoding corruption, duplicate rows, formula errors, impossible values, probable statistical outliers, and semantic inconsistencies between related fields.

Why it matters

For Excel users, the practical benefit is less time spent on repetitive cleanup before analysis. Clean data is not just a cosmetic issue. If the source table contains mixed spellings, text values that should be numbers, or broken identifiers, later work such as formulas, lookups, and reporting can produce misleading results.

For admins and managers, the feature is relevant because it signals how Microsoft is shaping Copilot in Excel around guided actions rather than only free-form prompting. For makers and power users, the value is in having a skill that can be invoked directly and that focuses on a common task they already know they need to do. The post specifically argues that Copilot Skills are useful when they reduce the time spent figuring out where a feature lives or how to prompt it.

The material does not say anything about licensing, tenant settings, or regional availability. It does say the feature is rolling out, so readers should assume it may not appear everywhere at once. That makes this more of a feature watch item than a change every organization can act on immediately.

How it works

The workflow described in the research notes keeps the user in control. First, you select the relevant data or table. Then you open Copilot in Excel and invoke the Clean Data skill, either through the skill name or by using the command path shown in the material. After that, Copilot inspects the data and identifies issues it can help correct.

From there, the user reviews what Copilot found and decides whether to accept the proposed corrections. The notes say Copilot explains why each issue matters, so the process is not just a blind cleanup action. That matters for spreadsheet work because some changes are safe to apply automatically, while others need review.

1. Select the table or range you want to check.
2. Open Copilot in Excel and call the Clean Data skill.
3. Review the issues Copilot identifies.
4. Accept or reject the proposed corrections.
5. Continue with formulas, PivotTables, or reporting after cleanup.

The examples in the material show why this can save time. Instead of manually fixing spacing, text casing, or text-formatted numbers one problem at a time, users can ask Copilot to find patterns and suggest corrections. That can be especially helpful before building summaries or other analysis that depends on consistent source data.

Limitations and open questions

The material is clear that the skill is rolling out, but it does not say whether it is available to all tenants, licenses, or regions. It also does not spell out which Excel versions are required, or whether the feature appears identically across web, desktop, and other clients.

There is also a difference between the source post and the research notes. The post emphasizes direct use of the skill and its broad list of problems, while the notes add a separate Clean Data command from the Data tab and a fuller workflow description. Because the source material is short, we cannot verify every interface detail from the post alone.

Finally, the material does not show how the skill handles edge cases. For example, it lists categories such as outliers and semantic inconsistencies, but it does not explain the exact rules Copilot uses to decide what to fix or how it presents alternatives when more than one correction is possible.

Sources

• Ami Diamond [MVP] video: New Clean Data Skill in Copilot Excel
• Ami Diamond [MVP] LinkedIn profile

Frequently asked questions

What is the new Clean Data skill in Copilot for Excel?

It is a Copilot skill in the Data & Analysis category that helps inspect and fix common spreadsheet quality issues. The post says it can be called directly with an @mention.

What kinds of problems can it find?

The material lists inconsistent capitalization and spacing, typos, numbers or dates stored as text, invalid identifiers, invisible characters, duplicate rows, formula errors, outliers, and related-field inconsistencies.

Is it available everywhere yet?

No availability details are given. The material says it is rolling out, so it may not be visible to every customer or region yet.