Their numbering, their order, every item
Quote each comment, say what you changed, and say where it now sits by section and line number. Comments you have not acted on are addressed rather than skipped, with a reason a reasonable person would accept.
The temptation is to group related comments for elegance. Resist it. An editor is checking that each numbered point was handled, and anything making that check harder works against you regardless of how well the manuscript now reads.
Disagreement is allowed and it has a form
Where a reviewer is wrong, say so respectfully, give the evidence, and offer something — a clarification in the text, an acknowledgement in the limitations. A flat refusal with no accommodation reads as defensive even when it is entirely correct.
Where two reviewers contradict each other, name the contradiction and explain which way you went. Editors expect to adjudicate that, and they cannot do it if your choice is silent.
Tone does more work than it should
Reviewers gave their time unpaid and the editor is reading for cooperativeness as well as substance. Brevity, specificity and an absence of irritation are worth more than most authors expect.
None of which means capitulating. The strongest letters disagree clearly on two or three points and comply straightforwardly on everything else, which is also what an honest revision usually looks like.
There is a practical reason to draft the letter before finishing the revisions rather than afterwards. Writing the responses forces you to decide what you actually think about each comment, and several of those decisions change what you do to the manuscript. Written afterwards it becomes a description of edits already committed to. It also means the manuscript and the letter cannot drift apart, which is the other way these submissions come unstuck. There is a second reason as well. A letter drafted first tells you how much work the revision actually is, which is the number you need before agreeing a resubmission date with an editor who has already set one.
Three steps, then the record keeps itself.
Send the letter and the manuscript
The full reviewer comments, the editor's letter and your paper as it currently stands.
Get the response drafted
Point by point in their numbering, each change located by section and line, disagreements argued rather than deflected.
Read it before it goes
You check anything you would not defend, and you submit it yourself.
Queries on record.
Do I have to accept every comment?
No, and a letter accepting everything uncritically is not necessarily strong. Where a reviewer has misread something, the usual cause is that your text was ambiguous, so the productive response is to clarify it and say so. Where they are simply wrong, argue it with evidence and offer a limitation acknowledgement.
How long should the response be?
As long as the comments require. Each item gets what it needs, sometimes a sentence and sometimes a paragraph with a table. What makes these letters unreadable is not length but structure, so keeping strictly to the reviewers' own numbering matters considerably more than being brief.
What if the reviewers contradict each other?
Name it explicitly and explain which way you went and why. This is common and editors expect to adjudicate. What they cannot do is guess which comment you followed, so a silent choice leaves them to discover that the other reviewer's point appears to have been ignored.
Should I show the changes?
Follow the journal's instruction, and where you have a choice supply both tracked and clean versions. Then locate every change in the letter as well, by section and line. An editor who can find each change in ten seconds is an editor giving you a fast and favourable read.
Is this like answering a committee?
The discipline is identical even though the setting differs. A dissertation committee, a review board and a journal all send numbered comments and all reward a reply that walks their list in their order. The habit transfers directly, which is why it is worth learning properly the first time it comes up.
Elsewhere in the hall.
Send the assignment and the deadline.
A specialist credentialed in your own field reads what you actually have and replies within two hours with a plan and a price. The reading costs nothing, and if the honest answer is that we are not the right hall for it, you will get that instead.
