Don't Trust the Counterparty's Redline: Verify Before You Rely On It

The counterparty sends back your draft with tracked changes. You open it in Word, walk through the Review pane, accept what you can live with, flag what you cannot, and send a response. This is the standard workflow. It is also where a specific, avoidable class of mistake lives: the change the counterparty made that was not tracked, and that you never saw.

Untracked changes are not usually malicious. They are almost always the result of a reviewer turning Track Changes off for a minute to fix formatting and forgetting to turn it back on. Or accepting a block of early edits to make the redline readable. Or pasting content from another document and having Word strip the tracking metadata during the paste. The effect on you is the same: you trust the tracked changes, miss the untracked ones, and either sign something you did not review or send back a response that ignores an edit the other side is counting on.

This post is about how that happens, why Word by itself is not enough to catch it, and the verification step that closes the gap.

How Word's tracking breaks

Track Changes is a document-level state. It is either on or off, and anything you type while it is off is permanent and silent. There is no warning when you save a document with Track Changes turned off, no indicator in the file that a span of text was edited without tracking, and no way — from the file alone — to tell whether tracking was ever disabled during the edit session.

The common ways tracking breaks during normal use:

• Toggled off for cleanup. A reviewer turns tracking off to fix a typo, adjust spacing, or re-number a list and forgets to turn it back on. Anything they edit afterward is untracked.
• Early accept. A reviewer accepts the first fifty tracked changes from a previous round to make the current round readable, then keeps working. The accepted changes are no longer tracked — but neither are any edits the reviewer made to the accepted text afterward, if they happen to have tracking off during the accept.
• Paste-and-merge. Copying content from another document and pasting it in. Depending on the paste method, the tracked-change state of the pasted content can be lost, accepted, or inconsistently preserved.
• "Final" view confusion. Word's Simple Markup and No Markup views hide tracked changes visually. Reviewers editing in these views can forget they are looking at a filtered version of the document and make edits that land outside the tracked-change scope.
• Macros and automation. Some add-ins and macros modify documents in ways that bypass tracking. The modifications appear in the saved file but not in the Review pane.

None of these are bugs in Word. They are consequences of Track Changes being a per-action feature rather than a per-document audit log. Word is faithfully recording the changes the user asked it to track and silently accepting the ones it was not asked to track. The file format has no concept of "this section was edited without tracking" because the file format treats tracked changes as an annotation layer over what is otherwise a normal document.

Why the Review pane does not catch it

The Review pane in Word shows you every tracked change in the document. That is exactly what it is supposed to do. What it cannot show you is a change that was never tracked in the first place, because there is no data in the file for the Review pane to read. The untracked edit is just part of the document text now. It is indistinguishable from the original.

This is the reason "just walk through the Review pane" is an incomplete verification workflow. It verifies the tracked changes. It does not verify that the tracked changes are the complete set of changes. The reviewer can be diligent, the Review pane can be exhaustive, and the document can still contain edits neither of them is aware of.

The independent comparison step

The fix is an independent comparison between the document you received and the last version you know is correct — typically the version you sent out or the last version both sides agreed on. A comparison surfaces every textual difference between the two files, regardless of whether the difference was tracked.

The mental model: the Review pane tells you what the other side intentionally flagged as a change. The comparison tells you what actually changed. The two should match. If the comparison finds differences that do not appear as tracked changes in the received document, those are untracked edits and they deserve your attention.

You can run a comparison with Word's built-in Compare feature (Review tab → Compare → Compare), or with a standalone tool. The standalone approach is typically less noisy — Word's Compare shows every character-level difference equally, including formatting drift, and the signal you are looking for can be buried.

Keeping a reliable baseline

An independent comparison is only as reliable as the baseline you compare against. If you do not know which version of the document you last approved, or if you have been overwriting the same working file round after round, you do not have a baseline and the comparison step does not work.

The minimum discipline:

• Save every round. When you send a draft out, save the version you sent with an unambiguous filename. Never overwrite it.
• Save every received version. When a draft comes back, save it with a separate filename. Do not merge it into the outgoing file.
• Name by direction and round. Something like Contract_v3_sent_2026-04-08.docx and Contract_v3_received_2026-04-10.docx. Leave no ambiguity about which was sent, which was received, and when.

With that discipline in place, the verification step is cheap: at the start of each review, compare the version you received against the version you last sent. Any difference that does not appear in the Review pane is an untracked edit.

When this matters most

Not every round of every deal needs independent verification. The workflow overhead is not worth it for low-stakes agreements with trusted counterparties. It is worth it when:

• The stakes are high. Any deal where a missed edit would cause material financial or legal harm — M&A, financing, major commercial agreements, IP assignments.
• The round is late. The closer you are to signing, the less room you have for "I assumed we were working from the agreed text." Verify the final few rounds even if you skipped the earlier ones.
• The document is long. Long contracts have more provisions where an untracked edit can hide. A 5-page NDA is easy to eyeball; a 60-page services agreement is not.
• You do not fully trust the workflow on the other side. This is not about trusting the counterparty personally. It is about whether their process is disciplined enough that you can rely on "they would have tracked it if they changed it."

A minimum-viable verification workflow

Here is a workflow that is cheap enough to actually run on every round that matters:

1. Save outgoing and incoming versions separately. Unambiguous filenames, no overwrites.
2. Open the received version. Walk the Review pane. Understand what the counterparty says they changed.
3. Run an independent comparison. Compare the received version against the version you last sent. Use a tool that surfaces material changes clearly and separates formatting from substance.
4. Reconcile. Every substantive difference in the comparison output should correspond to a tracked change in the received document. If not, the discrepancy is an untracked edit. Raise it.
5. Archive the comparison output. Save it alongside the round. If a question comes up later, you have a record of what was surfaced.

The comparison step adds a few minutes per round. It is the cheapest insurance you can buy against the failure mode of trusting the Review pane in isolation.

Frequently asked questions

Can I trust the redline the counterparty sends me?

You can trust the tracked changes that appear in the document. What you cannot assume is that those tracked changes reflect every change the counterparty made. Track Changes can be turned off mid-edit, changes can be accepted before the file is sent, and some operations (like paste-and-accept) strip tracking metadata. The only reliable way to know everything that changed is to run an independent comparison against your last known-good version.

How do I detect untracked changes in a Word document?

Compare the document you received against the version you last agreed on. An independent comparison — either Word's built-in Compare or a standalone tool — will surface every difference between the two files, including changes the counterparty made without tracking them. Any difference in the comparison output that is not reflected as a tracked change in the received document is an untracked edit.

Is this common or paranoid?

It is common enough to have a name in practitioner guides: "untracked changes" or "hidden edits." It is almost never malicious. It is usually the result of a reviewer toggling Track Changes off for a minute to clean up formatting and forgetting to turn it back on, or accepting a batch of early changes to make the redline easier to read. But the effect on you is the same: you review the tracked changes, miss the untracked ones, and sign a document with edits you did not see.

Does a comparison catch everything?

A comparison catches every textual difference between two documents. What a comparison cannot catch is a change between a version you never saw and the current version — if you skip a round, any edits in that gap are invisible. The verification workflow only works if you have a reliable baseline: the last version you actually reviewed and approved.

What does "redline provenance" mean?

Provenance means knowing where the redline came from and what it was compared against. A redline file by itself tells you what the markup says; it does not tell you what the unmarked baseline was, whether the comparison was run on the right two files, or whether anything happened to the document between the comparison and the version you received. Preserving provenance means keeping the baseline alongside the redline, so you can re-run the comparison yourself at any time.

---