# Remove metadata from an Excel workbook

> An Excel workbook can hold the author and last editor, a company name, threaded comments labeled with each commenter's account name, notes, the folder the file was last saved in (often including a Windows user name) and, for old shared workbooks, a log of user names. This tool removes those from .xlsx files, cuts the folder path of a linked file, or of another workbook named in a formula, down to the file name, leaves cell values alone, and lists what it can still find in the cleaned workbook, such as data connections.

## What an .xlsx file keeps about the people who worked on it

Spreadsheets travel. A budget goes to a client, a parts list goes to a supplier, a data export goes to a regulator. Each of those files can carry a surprising amount of history.

**Threaded comments.** Excel for Microsoft 365 has two kinds of annotations. Microsoft explains that threaded comments are labeled with the name of the person signed in to the account, and that this name cannot be changed by typing over it. A reply thread can reveal who on your side discussed a number, and what they said about it.

**Notes.** The yellow boxes that older versions called comments are now called notes. Excel labels them with whatever name is set in its options, which is often a full name.

**People list.** Workbooks with threaded comments include a list of the people who took part, with their identities.

**Saved location.** Excel records the folder the workbook was saved in. On Windows that path commonly runs through the user's profile folder, so it contains the user name.

**Shared-workbook history.** The legacy Shared Workbook feature keeps a revision log and a list of the user names who had the file open. Microsoft notes that the Document Inspector cannot remove comments, properties or personal information from a workbook while it is shared.

**Properties.** Author, last modified by, company, manager, title, dates, custom properties and any custom XML data a template added.

**Names in less obvious places.** A PivotTable records who refreshed it last and when. Scenario Manager stores the user who made each scenario. Custom views in a shared workbook are saved per person. A range that chosen people may edit without the password lists their accounts.

## How to clean an Excel workbook

1. Choose your .xlsx workbooks with **Choose files**, or drop them on the page.
2. Wait for each workbook's result. If the notice **Still in the cleaned copy** shows up first, read it before anything else.
3. In the **Before and after** table, check rows such as **Comments**, **Comment authors**, **Saved file location**, **Custom properties** and **Photos with metadata inside**.
4. Look at the **Removed** line under the table.
5. Download the cleaned workbook and open it once to check the sheets look right.

## What is removed

- document properties: author, last modified by, title, subject, keywords, dates and revision count;
- company, manager, template and total editing time;
- the program name and version, in the properties and in the workbook itself (Excel writes them again on every save);
- custom properties and custom XML data;
- threaded comments, notes, their authors and the people list;
- the shared-workbook revision log and its user-name list;
- the saved folder path;
- who last refreshed each PivotTable, and when;
- the user name and comment stored with Scenario Manager scenarios (the scenarios stay);
- custom views (**View > Custom Views**), in the workbook and in every sheet, because a shared workbook saves one under each person's name; set a view up again in the copy if you need it;
- the list of accounts allowed to edit a protected range without its password. A range that has a password keeps it, so those people need the password from now on. A range that never had one is removed altogether, which leaves its cells locked like the rest of the protected sheet. The result says so in words;
- Save as Web Page settings, which record where a sheet was published;
- sensitivity-label data and the list of link targets Excel keeps with the properties;
- the preview thumbnail, if one was saved;
- stored printer settings, which name a printer and can include its network path (page margins and paper size stay);
- metadata inside photos placed on a sheet, with the photos looking the same.

When a sheet showed its notes through a drawing that held nothing but those notes, the reference to that drawing goes too, so the cleaned file does not ask Excel for a part that no longer exists.

## Paths cut down to the file name

A link to another workbook is stored with its whole location, for instance `C:\Users\dana\Finance\rates.xlsx`. The tool keeps `rates.xlsx` and drops the folders before it. The same happens to the topic of a DDE link, to the file a data connection was imported from, and to a picture description that Excel filled with the picture's original path. Values already in your cells are not affected, but a link shortened like this looks for its file next to the workbook, so Excel may ask where the source is when you update links.

Formulas are treated the same way in one spot only: where a reference names another workbook. `'C:\Users\dana\Finance\[rates.xlsx]Sheet1'!A1` becomes `'[rates.xlsx]Sheet1'!A1`, whether it sits in a cell, a defined name, a validation rule, a conditional format, a table column or a chart range. Text between double quotes belongs to what the formula says, so it is left as typed even when it is a path, as in `=HYPERLINK("C:\Users\dana\plan.pdf")`. The result then lists that path as a folder path still in the copy.

This has a limit. A path is recognized only where it stands in one piece. If a formula builds a location by joining pieces of text, as in `="C:\Users\" & A1 & "\rates.xlsx"`, the tool cannot tell that the pieces add up to a path, and it changes nothing. A piece that looks like a path by itself, here `"C:\Users\"`, is still listed in the result; pieces that do not are not.

## Kept, and named in the result

These stay in the workbook, because removing them would break it or change what it shows. The **Still in the cleaned copy** box lists each one that is present and the part of the file that holds it:

- **Data connections.** The connection keeps its server, database, query and sometimes an account name, since the workbook needs them to refresh.
- **Links to other files and live data**: external workbooks, DDE links and other linked sources, after their paths were shortened, plus any link into a private network.
- **Digital signatures and signature lines.** A signature holds the signer's certificate and the signing time; a signature line holds the suggested signer's name, title and email address.
- **Mail links in cells and shapes**, counted one per link: a link that opens an e-mail message carries the recipient's address in the file.
- **Data stored by add-ins, macros and controls**, such as an add-in's saved settings, custom ribbon buttons and the properties of form controls.
- **Ink.** Handwriting and pen drawings on a sheet are content and stay; the box lists the ink data, which can hold stroke times and the pen or device used.
- **Unrecognized parts and markup.** The tool has no way to vouch for these, so it reports them.

Names that the workbook's properties record, for example its author, the person who saved it last or someone who wrote a comment, are searched for in the workbook's own text as well, such as its cells, sheet names and text boxes. Each one that turns up there goes into the box, visible or hidden alike, and you are the one to judge whether it should stay.

## What is not removed

Cleaning removes metadata. It does not judge or edit your data, so these stay exactly as you left them:

- **Hidden rows, columns and sheets.** Hiding is not deleting. Anyone can unhide them. Microsoft's Document Inspector can find and remove hidden rows, columns and worksheets; do that in Excel first if they hold anything you do not want to share.
- **Hidden shapes and text boxes.** An object that was hidden in the Selection Pane stays in the workbook. If it holds text, the result lists that as hidden text still in the file, with the drawing it belongs to, so you can find it in Excel. The stand-in shapes Excel keeps for buttons and other form controls are not counted, since their text is on the sheet.
- **Cells that look empty.** White text on a white cell, a number format such as `;;;` that shows nothing, a tiny font: the value is in the workbook and the tool treats it as an ordinary cell. Only rows, columns and sheets that are hidden, and hidden objects, are recognised as hidden.
- **Paths and internal addresses typed into cells.** A cell that holds something shaped like `D:\Team\plan.xlsx`, `\\server\share\…` or `http://intranet/…` is data and stays. The result adds a **Good to know** note with the count and the sheet, so you can decide. Other slash-separated text, such as the path half of a web address, is not mentioned.
- **PivotTable source data.** With the PivotTable option **Save source data with file** turned on, Microsoft notes that the PivotTable's source data is included in the workbook. A summary table can therefore carry every row it was built from; only the refresh name and date are taken out.
- **Defined names**, including hidden ones, and the formulas in your cells. One detail in them changes: where a name or a formula reaches into another workbook by its full location, the folders are dropped and the file name, sheet and cells stay, as described above. Two sorts of hidden name do go, because they belong to things listed under What is removed: the names Excel made for custom views and the ones the old Save as Web Page wizard left behind, in both cases only when no formula uses them.
- **Headers and footers**, where a name or a file path is sometimes printed.

Only .xlsx files are accepted. Workbooks with macros (.xlsm), binary workbooks (.xlsb) and the older .xls format are not supported; save a copy as .xlsx in Excel first. Files over 300 MB and password-protected workbooks cannot be read. Size is also checked after unpacking: the workbook's XML parts, sheets included, may total 256 MB (pictures are not counted), and a single part or picture may be 64 MB at most; beyond that the page says the workbook is too large instead of risking a crash. A part whose elements sit more than 512 levels inside one another gets the same message; a part read through in a single pass, such as the cells of a sheet, may go to 2,048 levels. Also turned away: a workbook package holding over 10,000 entries, and an element carrying more than 2,000 attributes, namespace declarations included. A part in which one piece of text goes beyond roughly 4 million characters is cleaned all the same and reported as not checked. The whole package, counting pictures and the parts the tool never opens, may unpack to no more than 512 MB. Pictures on a sheet are cleaned if they are JPEG, PNG, WebP or HEIC. Anything else that is embedded, for example a TIFF scan, an EMF drawing or an attached file, is left untouched and named in the result.

## What this tool cannot do

- **Work out a path that a formula builds.** As explained under Paths cut down to the file name, a location joined from pieces of text is not recognized; only a piece that reads as a path by itself is listed.
- **Read what a password hash encodes.** A protected sheet or workbook stores the hash and the salt of its password as encoded binary data. Well-formed values of the right length are left alone and nothing is read out of them; a value in those places that is not well-formed is checked like ordinary text and reported.
- **Look for a name that is a stock word.** Excel and Windows fill in defaults. If the recorded author or company is one of "microsoft", "test", "unknown", "guest", "author", "owner", "administrator", "admin" or "user" and nothing more (capital letters make no difference), it is cleared from the properties, and the tool leaves it at that: it does not hunt for the word in sheets, styles or names. The same goes for a recorded name that is no more than two Latin letters or digits, or that has no letter or digit at all (an id of zeros counts as that), with one difference: the result says that it was not searched for. Names in other scripts are searched for at any length.
- **Undo every coding a name can have.** When the tool looks for a recorded name or account in the cleaned workbook, it also reads the text with these spellings resolved: the `_x004A_` form Excel writes for unusual characters, character references, percent-escapes, a doubled quotation mark inside a formula, an apostrophe doubled inside the quotes around a sheet or workbook name (`'O''Brien'!A1`), `&&` standing for one ampersand in a page header or footer, `+` for a space in the query of a link, JSON escapes, and differences of capitals or invisible characters. That list is complete. The backslashes and quotes in a number format (`\J\a\n\e`), base64 and a domain in its `xn--` form are left undecoded, so a name hidden that way is not found.
- **Spot a folder path spelled in a rare way inside a longer text.** What is recognized: any value that is a path from start to end (a link target, a description, a property), with two reservations. On an old Mac disk that carries a custom name, the path has to finish in a file name and its extension, so `Work Disk:Profiles:Alice:budget.xlsx` is seen, while `Work Disk:Profiles:Alice:plan` is not. A capitalized `File:` needs a slash or a drive letter behind the colon, so `File:Users` is not seen either. Recognized inside longer text: a drive letter and a slash, a share such as `\\nas\finance`, `file:` addresses, the `/home` and `/Users` family, `~/`, an environment variable such as `%TEMP%\…`, `Macintosh HD:…`. Possibly missed inside longer text: forms like "C:x\Alice\notes", where no slash follows the drive letter, and a bare root with no name behind it, like "~jane/" in a sentence.
- **Get rid of a cropped picture's hidden edges.** When a picture on a sheet is cropped in Excel, the full image stays in the workbook and the crop can be reversed by whoever opens it. The tool cleans picture metadata but leaves the pixels, so the cut-off area remains. To remove it in Excel, pick the picture, open **Compress Pictures** from the **Picture Format** tab and turn on **Delete cropped areas of pictures**; inserting a saved copy of the cropped picture in its place does the same.
- **Outwit a workbook built to fool it.** Someone who crafts an .xlsx on purpose to get data past the checks can use a few constructions the tool misreads, such as numbers or labels written in forms Excel itself never saves, or parts that contradict each other. A name in such a workbook can then stay without being listed. Content repeated many thousands of times can also make processing take noticeably longer.

Read an empty result as "nothing this tool checks for was found". The [limits section of the methodology page](/methodology/#limits) lists what that leaves out.

## A short check before you send a workbook

Cleaning deals with the records listed above. A few checks in Excel cover the rest:

1. Right-click any sheet tab. If **Unhide** is available, at least one sheet is hidden; unhide it and decide whether it should go.
2. Open **Formulas > Name Manager** and look for names that point to other files or to data you meant to remove.
3. Look for PivotTables and decide whether their source data should travel with the file.
4. Print preview one page to see what the header and footer say.
5. Then clean the workbook here and read the result, which shows what was taken out and names the things it still found in the copy.

## Where your workbook goes

The workbook is unzipped and edited inside the browser tab you are looking at. Cell values, comment text and the names found in the file are not sent to any server, and nothing remains once you close the page. That makes it suitable for financial or client data you are not allowed to upload.

## How the Excel cleaning works

An .xlsx file is a zip package of XML parts. In a background worker, the tool opens the package with fflate, a compression library fetched from this site when your first Office file is cleaned. From the package's relationships it finds the comment, threaded-comment, person and revision parts, wherever in the package they are stored, and deletes them. Note boxes are taken out of their drawing parts, and a drawing part left with nothing in it goes too, together with the relationships that pointed to it. The saved-path element is removed from the workbook part, and the author, company, dates and link targets are cleared out of the property parts. Pictures placed on sheets are passed to the photo cleaner, which strips their metadata without touching pixels. Cell values and styles are copied unchanged, and so are formulas apart from the folders in a reference to another workbook, before the workbook is repackaged, with every zip entry dated 1 January 1980.

The repackaged file is then examined by a separate checker that does not reuse the cleaner's own lists. Working through the workbook one part at a time, it compares each relationship, element and attribute with the published definitions of the spreadsheet format. A spot the definitions reserve for a name, an account id, a date of editing, a location or a connection is reported when it still has a value, and so is anything the definitions do not describe. Only what is known and found clean passes; the rest goes into the result's list. To see how the tools are tested and how you can confirm that no upload happens, visit the [methodology page](/methodology/).

## FAQ

### Will my formulas still work after cleaning?

Formulas that stay inside the workbook are not modified, and neither are cell values or formats. A formula or a chart that reads from another workbook keeps the values it last showed, but its reference now holds only that workbook's file name, so Excel looks for the file in the same folder when you update links. Text between quotes in a formula stays exactly as you typed it.

### Why can't I remove my name from a threaded comment in Excel itself?

Microsoft ties threaded comments to the signed-in account, and its help page says you cannot remove or change that name. Deleting the comment, or cleaning the file here, removes it.

### Are notes removed as well as threaded comments?

Yes. Both kinds of annotation and their author names are removed.

### Does this unhide or delete hidden sheets?

No. Hidden sheets, rows and columns are left as they are. Check them in Excel before sharing.

### What is the "saved file location" row?

It is the folder path Excel stored when the workbook was last saved, such as a path inside a user profile. It is removed. Folder paths found elsewhere are shortened to the file name: in links to other workbooks, in formulas that point into another workbook, and in a picture description that is a path and nothing more.

### What happens to my data connections?

They stay, because a query or a refresh would stop working without them, and the result lists them under **Still in the cleaned copy**. A connection string can name a server, a database and an account. If the recipient should not see that, delete the connection in Excel (**Data > Queries & Connections**) and clean the workbook again.

### Can I clean a CSV file here?

No. A CSV file is plain text with no hidden properties. What you see in a text editor is everything it contains.

## Sources

Checked 2026-10-11.

- [Microsoft Support: The difference between threaded comments and notes](https://support.microsoft.com/en-us/office/the-difference-between-threaded-comments-and-notes-75a51eec-4092-42ab-abf8-7669077b7be3)
- [Microsoft Support: Insert comments and notes in Excel](https://support.microsoft.com/en-us/office/insert-comments-and-notes-in-excel-bdcc9f5d-38e2-45b4-9a92-0b2b5c7bf6f8)
- [Document Inspector categories for Excel workbooks (Microsoft)](https://support.microsoft.com/en-gb/office/remove-hidden-data-and-personal-information-by-inspecting-documents-presentations-or-workbooks-356b7b5d-77af-44fe-a07f-9aa4d085966f)
- [Microsoft Support: Design the layout and format of a PivotTable](https://support.microsoft.com/en-gb/office/design-the-layout-and-format-of-a-pivottable-a9600265-95bf-4900-868e-641133c05a80)

Last updated: 2026-10-17