A document with four reviewers and three rounds of tracked changes is not a reviewed document. It is an archaeology site, and the writer is expected to reconstruct the original building from the layers.
The tools are not the problem. Track changes, comments, and suggestion modes are all good tools. The problem is that nobody agreed on who does what, in which order, and what "done" looks like. A workflow fixes that in about ten minutes of setup.
Decide the roles before the first comment
Every review needs three roles, even if one person holds two of them.
The author owns the document and makes the final call on every change.
The reviewers comment and suggest. They do not resolve their own comments or accept their own changes.
The editor (sometimes the author) resolves comments, accepts or rejects changes, and produces the clean version.
In a two-person review, the author is also the editor and the other person reviews. In a larger review, someone has to be appointed as editor or the document will sit with fifty open comments until the deadline passes.
Write the roles into the email that sends the draft. One line is enough: "Sam and Priya are reviewing, I will resolve comments and send the clean version Thursday."
Set the mode
Most word processors offer at least three modes: editing, suggesting, and viewing. Reviewers should work in suggesting mode, which records every change as a proposal rather than a fact.
This is not a matter of etiquette. A reviewer making direct edits to a shared file can silently overwrite the author's work, and the author may not notice until the document is published. Suggesting mode makes every change visible and reversible.
The author should work in editing mode, but only after the review closes. During the review, the author responds in comments rather than making changes, or the reviewers will be commenting on a version that no longer exists.
Make comments actionable
A comment that says "unclear" gives the author nothing to do except guess. A comment that says "I am not sure what 'the second option' refers to — the pricing model or the delivery date?" gives them a decision.
The shape of a useful comment is the same as useful feedback anywhere: quote the thing you are responding to, describe what you experienced as a reader, and ask a question if you have one.
Use comment threads for discussion and keep them in one place. If a topic needs more than three replies, it has outgrown the comment box and should become a short meeting or a separate document.
Resolve, don't accumulate
The most common failure in a long review is the comment that is answered but never closed. Six months later the document carries a hundred resolved and unresolved comments in the same visual state, and nobody can tell which decisions were made.
Adopt a rule: a comment is resolved by the editor, not the reviewer, and only after the author has responded. The reviewer's job ends when they have raised the issue. The author's job ends when they have decided. The editor's job is to record the decision.
This also prevents the passive-aggressive loop where a reviewer reopens a comment because they did not like the answer. If the author has responded and decided, the comment is resolved; if the reviewer wants to escalate, they take it out of the document and talk to the author directly.
Three rounds is usually enough
Structured reviews tend to follow a predictable shape.
Round one: structure. Does the argument work? Is anything missing? Reviewers read the whole document once and comment at the section level. No line edits.
Round two: detail. Paragraphs and sentences. Reviewers suggest specific changes. The author accepts, rejects, or replies.
Round three: proof. Spelling, punctuation, formatting, consistency. Ideally a different person from rounds one and two, because the earlier reviewers have stopped seeing the text.
Trying to do all three at once produces a document where a comma correction sits beside a comment that says "the whole second half should be cut." Both may be right, but the writer cannot act on them together.
Keep a clean version
At the end of every round, the editor should produce two files: a tracked version with all changes visible and a clean version with everything accepted.
The clean version is the one people read. The tracked version is the record. Keeping both means the author can always answer the question "what changed in this round?" without opening three files and comparing them by eye.
For text that has moved between tools, the text diff shows exactly what changed between two versions, including changes that a word processor's tracking might miss after a copy-paste.
Closing the review
A review is closed when four things are true.
- Every comment is resolved or explicitly deferred with a name and a date.
- Every tracked change is accepted or rejected.
- A clean version exists and has been read once from the beginning.
- The document has a version number or date in its name.
Announce the close in writing, even if it is one line. "Review closed as of Thursday; the clean version is attached; thank you both." This is the signal that stops people sending comments on Friday.
When the review keeps reopening
If a document has been through five rounds and is still moving, the problem is usually not the document. It is one of three things: the brief was unclear, the decision-maker is not in the room, or two people disagree about something that the review cannot settle.
Stop reviewing and settle the disagreement directly. A ten-minute conversation often saves a week of tracked changes, and the document will be better for having a decision in it rather than a compromise between four people who never spoke.
The tools are there to record decisions, not to make them. A clean workflow keeps the record accurate and the decisions visible, which is the most any document can do.