A reviewed Word document is a list of decisions wearing a document's clothes. FindUtils DOCX Comments Extractor reads every review comment out of a .docx and hands you the list — author, date, the passage it points at, the comment, and the comment it replies to — as CSV or a Markdown table, in your browser, without modifying the file.
This guide explains where comments actually live inside a .docx, what the columns mean, and the two limits worth knowing before you rely on the export.
Where Comments Live Inside a .docx
A .docx is a zip. Rename one to .zip, open it, and the review pane turns out to be three separate parts:
word/comments.xml— the comments themselves: an id, the author, their initials, the date, and the paragraphs of the comment body.word/commentsExtended.xml— Word 2013 and later only. Which comment is a reply to which, and whether a thread has been marked resolved.word/document.xml—commentRangeStartandcommentRangeEndmarkers wrapped around the passage each comment is anchored to.
The extractor reads all three, which is why the export carries the anchored sentence and the thread shape rather than just a pile of comment text.
There is one detail in there worth knowing, because it is where naive exports go wrong. commentsExtended.xml does not reference comment ids. It references the paraId of the last paragraph in each comment. A reply is linked to its parent through that paragraph, not through the comment number — so an exporter that assumes "comment 2 replies to comment 1 because 2 follows 1" will be wrong on any document where a reviewer answered an earlier thread late.
What the Columns Mean
| Column | What it holds |
|---|---|
id | Word's own comment number |
author | The reviewer's display name |
initials | The initials Word shows in the margin |
date | When the comment was written, as Word recorded it |
anchored_text | The passage in the document the comment points at |
comment | The comment body, with its own line breaks preserved |
parent_id | The comment this one replies to, empty if it starts a thread |
done | Whether the thread is marked resolved |
The anchored_text column is the one that makes the export readable on its own. A comment that says "this contradicts clause 4" is meaningless in a spreadsheet next to nothing; next to the sentence it was attached to, it is a task.
Comments Are Not Tracked Changes
This is the limit most likely to catch you out. Tracked changes are a different feature — insertions, deletions and formatting revisions live elsewhere in the document, and this tool does not read them. A document that came back with edits rather than comments will produce an empty result, correctly.
The empty result is deliberately not an error. A .docx with nothing in the Review pane is simply a .docx with no comments part inside it, and the page says so in plain words rather than showing a red failure that sends you looking for a problem that does not exist.
When Every Row Says "Open"
If every comment shows as unresolved, the document probably came from an older Word or from another word processor. The resolved flag lives in commentsExtended.xml, which Word 2013 introduced — a file written without that part has no resolved information in it at all.
The comments themselves still come out. Only the reply links and the resolved column depend on that part being present.
Using It
- Drop the reviewed
.docxonto the page, or click to choose it. - Read the table. Replies sit directly under the comment they answer, indented, rather than in document order.
- Pick CSV for a spreadsheet or Markdown table for a ticket, a pull request description or a wiki page.
- Download. The file is named after the document, so
contract-v7.docxgives youcontract-v7-comments.csv.
The CSV is plain RFC 4180 — a header row and then the data, with commas, quotes and line breaks inside a comment quoted properly. There is no banner or comment line above the header, so it opens in Excel, Numbers or Sheets without an import step.
What This Is Actually For
Three situations come up repeatedly:
A review round that needs to become a task list. Thirty margin comments become thirty rows with an author and an anchor, which can be sorted, assigned and ticked off. Sorted by author, it also shows who raised what and which comments are replies rather than new points.
A contract or compliance markup that needs a record. Every point addressed, and evidence that it was. A table of clause text, reviewer and comment is that record, and it survives the document being superseded.
A file you cannot open. On a machine without Word, the comments in a .docx are effectively invisible. Reading them out of the zip does not require the application that wrote them.
It Only Reads
The document is never written to. No comment is removed, resolved or edited, and no new .docx is produced — the only download is the comment table. If you want to strip the properties a Word file carries about its authors, that is a different job and a different tool: DOCX Metadata Remover reports and clears the author, last editor, company and template, and rebuilds the file.
If you want the document's body text rather than its comments, DOCX to Markdown and DOCX to HTML convert the content and leave the comments behind.
Limits
- Comments only. No tracked changes, no revisions, no formatting history.
- Reply links and the resolved flag need
commentsExtended.xml; without it those columns are empty. - A comment whose author is blank usually came from a file whose reviewer names were stripped on purpose.
- An older binary
.docis refused with a message rather than a broken result. Save it as.docxin Word first. - One document at a time.
Privacy
The .docx is opened in your browser and the parts inside it are read there. Nothing is uploaded, nothing is stored, and the file is not modified.
This one matters more than most. Review comments are candid by nature — they are frequently about people, and they are usually attached to a contract, a manuscript or a strategy document that is not public. Handing that file to an unknown website to find out what is in it is the opposite of what the job needs. You can confirm nothing leaves the tab in your browser's network panel.
