Rebuttal Week Playbook: The Day-by-Day Response Strategy That Flips Rejections

1. Day 0 — Read Everything, Respond to Nothing

When reviews land, the urge is to start drafting responses immediately.

Don't. Read every review twice without opening a text editor. Wait 24 hours

before writing a single word. Emotional replies — even polite ones — read

differently than calm ones, and reviewers can tell.

During this 24 hours, build a table. Rows = individual criticisms. Columns =

reviewer ID, criticism summary, severity (1-3), whether you agree, what you

can do about it in 5 days. This table is the rebuttal. Everything else is

just formatting.

2. Day 1 — Categorize by Severity

Every reviewer criticism falls into one of three buckets: factual errors in

their reading of your paper, methodological gaps you can address with new

experiments, and subjective calls where you disagree. Handle them differently.

Factual errors: quote the original paper text, explain where they misread.

Never say "the reviewer is wrong" — say "as stated in Section 3.2, we

specifically evaluate on..." The tone difference wins meta-reviewers.

Methodological gaps: decide in the first 48 hours whether you can run

the missing experiment. If yes, start it now. Baseline runs on GPU clusters

routinely take 72 hours — if you start on day 4, you won't have results.

3. Day 2-3 — Run the Experiments You Decided to Run

This is the hardest 48 hours. You committed to running experiments, and now

you need results, plots, and a one-paragraph summary — all before day 4.

Prioritize experiments that address Severity 3 criticisms from whichever

reviewer has the lowest score. Flipping one low-scoring reviewer up by one

point often changes the meta-reviewer's average enough to change the outcome.

If your first experiment fails (the baseline you added is actually better),

that's information. Don't hide it. Report it honestly and explain what it

reveals about your method's regime of applicability. Reviewers respond

much better to "we ran this, it showed X, which means our contribution

is specifically in regime Y" than to silence.

4. Day 4 — Draft, Aggressively

First draft: write responses in the order reviewers will read them, not

the order you want to argue them. That means most reviews go R1 then R2

then R3, and meta-reviewer reads top to bottom. Put the strongest positive

signals in R1's response. Put your new experimental results wherever they

address the highest-severity concern.

Response length: 350-500 words per reviewer is typical. Beyond that,

reviewers start skimming. If you have a 900-word response, cut the

defensive framing first, then cut anything that just restates their

concern before answering it.

5. Day 5 — The Three Patterns That Flip Decisions

Papers that move from borderline reject to accept tend to share three

response patterns. Use all three if you can:

Pattern 1 — The "concede and contain": explicitly agree with one of the

reviewer's criticisms, then narrow the scope of your claim to match. This

is powerful because it shows editorial integrity, and narrowed claims are

easier to accept than broad ones.

Pattern 2 — The "new evidence": a clean experiment with a clean result

that directly answers one Severity 3 concern. Present the result in a

single sentence and a small table. Do not re-litigate the original claim.

Pattern 3 — The "revised figure": replace one figure from the original

paper with a clearer version that reviewers can see in the rebuttal PDF.

A visual change is perceived as substantive progress even when the

underlying numbers didn't move much.

6. Day 6 — Advisor Review and Polish

Send the full rebuttal draft to your advisor at the start of day 6,

not the end. They need time to read, and you need time to act on feedback.

If your advisor is unreachable, send to a senior coauthor or a recent

PhD graduate in your group.

Common polish items: remove "we would like to thank the reviewer for"

prefixes (reviewers find them performative by 2026), replace "clearly"

and "obviously" with actual evidence, make sure every claim in the

rebuttal has a pointer to a specific section, line, or new result.

7. Day 7 — Submit, Then Disengage

Upload the rebuttal 6+ hours before the system deadline. OpenReview

and similar platforms have uploaded corruption at peak hours —

don't let a last-minute network timeout destroy a week of work.

Once submitted, close the tab. Do not refresh reviewer scores. Reviewers

have their own post-rebuttal discussion window; what you do in that window

is wait. Checking compulsively hurts your next paper — you need the

cognitive space back to start the next project.