You have a decision letter open in one tab, a marked-up manuscript in another, and a growing list of comments that do not map cleanly onto a single revision. Before you touch the draft, stop and build the artifact that will carry the whole revision: a response matrix. It is not a bureaucratic extra. For a course paper, thesis chapter, or first manuscript, the matrix turns scattered reviewer comments, supervisor notes, and your own flagged sections into one inspectable plan where every point either changes the manuscript or is answered in writing.
This guide walks through that process from the raw decision letter to the final check, including what to record, how to decide which comments are truly connected, and what completion looks like before you resubmit. The worked example uses a thesis chapter, but the same table works for a coursework draft returned with supervisor feedback or a January paper you want to restart cleanly.
What state are you starting from?
Before building the matrix, name your starting material. Three common states call for slightly different first steps.
You have a single reviewer report or supervisor memo. Copy each numbered comment into a sheet as its own row. If comments are unnumbered, split them at the author-facing instruction level: one row per requested change, clarification, or question.
You have multiple reviewer reports plus an editor's summary. Create one master list and add a column for Source so two reviewers raising the same statistical concern stay visible as separate entries. The same fix often answers both; the matrix should show that pairing rather than hide it.
You have nested feedback on a thesis chapter that also affects other chapters. Record the comment once, then add a column like Affected sections so a methods rewrite carries into results and discussion instead of being treated as a local edit.
Starting from a diagnosis matters because the matrix is a decision tool, not a transcription exercise. If you capture raw comments without marking the ones that require a choice — change the analysis, add a limitation, dispute the claim, postpone to future work — you end up with a list rather than a plan.
Which comments belong in the matrix at all?
Not every sentence in a review is actionable. COPE frames peer review as depending on trust, and part of that trust is treating the reviewer's assessment as constructive material to be engaged with seriously. That still leaves a sorting step.
Sort each review item into one of four categories at the top of your matrix:
| Category | What it means | What it produces |
|---|---|---|
| Change | You accept the point and will modify the manuscript | A tracked edit and a response sentence summarizing the change |
| Clarify | The science may be unchanged but the writing or reporting is unclear | A revised passage, figure legend, or method description |
| Answer | The reviewer asks a question that the manuscript already addresses | A short response that points to the exact location or explains the reasoning |
| Disagree | You do not accept the requested change | A respectful, evidence-based response that either explains the original design or proposes a middle path |
A common beginner error is treating every comment as a Change. That produces a revision that quietly adopts suggestions the advisor may not have meant as requirements, and it hides real disagreements until the resubmission letter. ICMJE guidance similarly points authors to the selected journal's specific instructions for what a submission must include, which is a reminder that journal-specific requirements sit alongside reviewer comments and should be checked separately, not merged into reviewer rows.
How should the matrix be structured?
The workable minimum is six columns, plus a source column when you have multiple reviewers. Do not over-engineer before you have processed ten comments; you will know by then if you need an extra column for cross-chapter impact or a co-author assignment.
Core columns
- ID — stable row identifier such as
R1-04(Reviewer 1, comment 4). This lets you cite rows in team discussion without restating the text. - Original comment — verbatim or lightly compressed. Keep the reviewer's own wording so your response can quote the exact concern.
- Category — one of Change, Clarify, Answer, or Disagree from the table above.
- Action — the concrete next edit or response, written as an instruction you could hand to a collaborator.
- Location — manuscript section, figure, or table where the change lands.
- Status — Not started, In progress, Done, Reviewed.
When the project needs more
- Add
Sourcewhen combining multiple reviewer reports. - Add
Affectswhen a thesis chapter change ripples into other chapters. - Add
OwnerandDue datewhen the revision is collaborative or tied to a submission deadline. - Add
Response textwhen you are drafting the author-response letter and want every row to include the exact sentence the editor will read.
A matrix works much like a project board once every row has a status. If you are managing a full paper revision alongside new idea exploration, WisPaper's Research Projects area gives you a place to keep the revision as one project among others, with the matrix as its central working file.
What does a response actually say?
A response row has a compact shape, especially for the Change and Disagree categories that editors read most carefully.
For an accepted change, write one or two sentences: what you changed, where it appears, and a brief statement of the new result if relevant. Resist the instinct to recap the entire manuscript.
For a disagreement, show the reviewer's reasoning was considered, state your position in the manuscript's own terms, and offer an alternative that respects both positions where one exists. NIH's rigor guidance reinforces that transparent design, reporting, and interpretation are part of a sound scientific method — when you are defending a design choice, the strongest version of that defense points to precisely those transparent reporting elements rather than to assertion.
Here is a worked example from a thesis chapter, with the author's private decision notes in the method shown but not in the final response letter.
Worked example: thesis chapter revision
Reviewer 1, Comment 3 (original): "The exclusion of participants with missing covariate data is not clearly justified. This could introduce selection bias."
| Field | Entry |
|---|---|
| ID | R1-03 |
| Source | Reviewer 1 |
| Original comment | Exclusion of participants with missing covariate data not clearly justified; possible selection bias |
| Category | Change |
| Action | Add a sensitivity analysis comparing excluded vs. included participants on key baseline variables; report results in Methods and Results |
| Location | Methods 4.2, Results 5.4, supplementary table |
| Status | Done |
| Response text | "We agree that the exclusion rule needed clearer justification. We now report, in Table S3, baseline characteristics of excluded versus included participants, and we add a sensitivity analysis in Results 5.4 showing that the main estimate is stable when the exclusion rule is relaxed. We also state this residual limitation explicitly in the Discussion." |
The response text is the artifact that leaves the working matrix and enters the author-response letter. Notice it contains the change, the location, and the honest remaining limitation without inflating the claim.
A second, smaller example:
Reviewer 2, Comment 7 (original): "The authors overstate the causal interpretation in the abstract."
| Field | Entry |
|---|---|
| ID | R2-07 |
| Source | Reviewer 2 |
| Original comment | Causal interpretation overstated in abstract |
| Category | Change |
| Action | Revise abstract to use associational language; align with Results wording |
| Location | Abstract |
| Status | In progress |
| Response text | "We have revised the abstract to describe the observed associations without causal language, consistent with the observational design and with the Results section." |
These response patterns are reused across categories. A Clarify entry often reads like a shorter Change; an Answer entry points to the location and adds one explanatory sentence; a Disagree entry shows the counter-argument and ends with the smallest compromise you can genuinely accept.
What checkpoints does the revision need?
Set checkpoints before you start editing. Without them, the matrix reports individual row status but cannot tell you whether the revision as a whole is converging.
Checkpoint 1: Every comment has a category and an action. No Not started row should exist without a written action by the time you finish the first pass through the decision letter.
Checkpoint 2: Every manuscript change is traceable to a row. If you edit the draft and cannot point to the matrix row that motivated it, either the edit is out of scope or the matrix is incomplete.
Checkpoint 3: Linked comments are resolved together. Two reviewers raising the same point should be answered in paired rows, with one shared location and response logic, so the response letter shows a consistent story.
Checkpoint 4: The response letter exists and quotes long comments. Editors read the letter alongside the tracked manuscript. It should contain the full original comment for each major point, not just your reply.
Checkpoint 5: The status column is honest. Done means the edit and the response text are both written and checked, not that you made an unlabelled change somewhere in the draft.
For a January reset use case, run this same checkpoint sequence against the previous semester's draft. The matrix becomes the backlog: unchanged rows become the first actions of the new writing block, and rows marked Done are the evidence that last term's work is actually finished rather than assumed finished.
How do you know the matrix is complete?
Completion is testable. A matrix is finished when all five of these are true together.
- Every numbered comment from every source appears in at least one row.
- Every row has a category, an action, a location, and a final status — no blank decision cells.
- Every
Donerow has corresponding text in the manuscript and a response sentence in the letter. - Every cross-referenced comment (reviewer to reviewer, chapter to chapter) is linked and answered once with a consistent position.
- A final read of the tracked manuscript finds no unexplained change; every edit maps back to a row.
The last test is the most diagnostic. If you cannot trace a change to a reviewer comment or to a journal-specific requirement you recorded separately, you have scope creep — either the change is redundant or the matrix missed a comment.
Once rows are complete, sense-check the references that support your responses. For papers that rely heavily on a reference list, TrueCite can screen your BibTeX file against real academic databases and flag entries that may be hallucinated or malformed; this is relevant when your response letter or added analyses cite newly introduced work. Run it as a final verification step once the manuscript text is stable, not as a substitute for reading the sources themselves.
Reusable checklist
- Extract every comment into a row; number rows by reviewer and comment ID
- Add a
Sourcecolumn when more than one reviewer or advisor commented - Assign one of Change, Clarify, Answer, Disagree to every row
- Write a concrete action and a manuscript location for every row
- Pair overlapping comments from different reviewers and cite the shared fix
- Draft response text for accepted changes, clarifications, answers, and disagreements
- Add
Affectsentries for thesis changes that propagate across chapters - Run the five checkpoints before any deadline discussion
- Confirm every manuscript edit maps to a matrix row
- Check the final reference list for entry integrity before submission
What is the next concrete action?
If you are standing in front of an unprocessed decision letter, do this in the next hour:
- Open a fresh spreadsheet or a new project board.
- Paste in the six core columns and the four-category sorting table.
- Transcribe the first ten comments, one per row, with verbatim or nearly verbatim wording.
- Sort them into Change, Clarify, Answer, or Disagree.
- Fill in the action and location for just those ten rows.
- Stop and review: if the categories feel forced, add a seventh column for your own note before continuing.
That small, finished slice gives you a template and a judgment test before the full list arrives. If your reference list is part of the revision and you expect many new citations, start gathering BibTeX entries as you read so the final screening step in TrueCite is possible without a last-minute export scramble.
For related papers you need to cite in your revision responses, WisPaper's search interface can help you find academic papers to support a new sensitivity analysis or a disagreement, and you can save the ones that matter to My Library so the working set stays together. Around the project itself, Research Projects keeps the matrix, the manuscript drafts, and the decision letter in one visible workspace rather than split across three machines. If a reviewer comment requires you to defend a preserved analysis choice with more background papers, you can also ask questions across the working set in Library QA once those sources are saved in My Library.




