How to get feedback on a research paper

Getting useful feedback on a research paper is one of the harder skills to develop in academic life — not because readers are scarce, but because good feedback requires the right reader, at the right stage, with the right question. Ask too early and you waste goodwill on a draft that will change entirely. Ask too late and there is no time to act. Ask without framing the request and you receive vague reassurance instead of substantive critique.

This guide covers the full arc: how to prepare before you send anything out, how to choose between different sources of feedback, how to brief a reader so they give you what you actually need, and how to act on the response without surrendering your voice in the process. There is also a section on where AI-assisted review fits alongside human readers — an honest comparison, not a sales pitch.

Before you ask anyone: the self-review checklist

Sending a draft for feedback before you have done your own critical pass is a common mistake. It strains your reader’s patience on problems you could have caught yourself, and it tends to produce surface-level comments rather than the deeper engagement you need.

Work through these questions before the draft leaves your hands.

Argument and structure

• Is there a clear, falsifiable claim that the paper is trying to establish or advance?
• Does the structure follow the logic of the argument, or did it follow the logic of how you did the research?
• Does each section do one job, and does it end before the next section’s job begins?

Literature and framing

• Have you positioned the work relative to the conversation it is entering — not just cited the papers you used?
• Are the gaps or tensions in the existing literature stated explicitly, rather than implied?

Methods and evidence

• Are your methodological choices justified, not just described?
• Is every claim in the results section supported by the data presented in the same paper?

Writing

• Read the introduction and conclusion back to back. Do they agree about what the paper does?
• Cut the first paragraph of the introduction. Does the paper improve? (It often does.)
• Are there any sentences that took more than two readings to parse?

Reviewing your own work this way, before sharing it, also makes you a better recipient of feedback later. You already know where the weak joints are, so you can calibrate how much weight to give a reader’s comments on those sections.

For more on what the self-revision process looks like in practice, see our guide to academic writing.

How to brief your reader well

A reader who does not know what you need will default to whatever kind of feedback they are comfortable giving. Some readers red-pen grammar; some reframe your argument in their own terms; some validate. None of this is necessarily useful.

A good brief takes two minutes to write and changes the quality of feedback you get entirely. It should cover:

Where you are in the process. “This is a first full draft” signals something different from “this is close to submission.” Readers calibrate depth and tone to that.

What you are most uncertain about. Be specific: “I’m not sure the theoretical framework in section two does enough work” produces better critique than “please give me feedback.” Name the sections or claims you most want scrutinised.

What you do not need right now. If the journal has language-editing services and copyediting will happen before submission, say so. This frees your reader to engage with substance.

The timeline. Tell them when you need the response, and be honest about how firm that is. Academic time is already stretched; a clear deadline is a courtesy.

One sentence on each of these points is sufficient. The brief does not need to justify the paper or apologise for its state — just orient the reader to where they can do the most good.

Sources of feedback, and what to expect from each

Your advisor or supervisor

For doctoral students and postdocs, the supervisor is usually the first and most important reader — and also the one whose feedback is hardest to interpret. They know the field and your specific project, but they are also assessing your development as a researcher, which means their comments sometimes address the writer as much as the writing.

What advisors do well: identifying gaps in the literature, catching methodological problems, flagging claims that the field will challenge.

What to watch: advisors sometimes rewrite papers in their own voice, or push you toward their theoretical preferences. Learn to distinguish feedback that strengthens your argument from feedback that substitutes their argument for yours.

Co-authors and collaborators

Co-authors bring accountability that a reader-as-favour cannot: they have a stake in the outcome. This tends to produce thorough, honest critique.

The risk is the opposite of an advisor’s: co-authors may be reluctant to raise concerns that would create conflict, especially about shared decisions already made. It helps to explicitly invite critique of the parts you made together, not just the parts you drafted alone.

Peers in your field

A colleague at roughly the same career stage, working in an adjacent area, can be the most useful reader you have. They are close enough to the literature to catch technical errors, but far enough from your specific argument that they will flag assumptions you have forgotten to explain.

The practical challenge is reciprocity. Peer review among colleagues works best as an explicit exchange — you read mine, I read yours — and that requires coordination that busy people do not always manage. Writing groups and informal reading circles at universities exist partly to solve this.

Writing centers and editorial services

University writing centers are under-used by researchers, in part because researchers assume they are for undergraduates. Most are not. Many offer consultations on scientific and academic writing at the postgraduate level and above, and skilled writing tutors can identify structural problems that field-specific readers normalise.

Professional academic editing services occupy the same space with faster turnaround and broader availability. They are most useful for clarity and flow rather than argument — a good editor will not tell you whether your methodology is sound, but they will tell you whether your reader can follow it.

AI peer review

AI tools now represent a genuinely distinct option — not a replacement for human readers, but a source of feedback with properties that human readers cannot match.

The practical advantages are availability (you can get feedback at 11pm before a submission deadline), consistency (the tool applies the same analytical framework across the whole paper), and speed (results in minutes rather than weeks). A well-designed AI reviewer can identify structural gaps in the argument, note where claims outrun their evidence, flag undefined terms, and surface literature that the paper may be missing — the category of feedback that peer reviewers and editors actually spend most of their time on.

The honest limitation is that AI lacks the kind of expert judgment that comes from years of working in a particular subfield. It cannot assess whether a theoretical claim is genuinely novel or whether a methodological choice is reasonable given community norms you have not made explicit. It can tell you that a section is unclear; it cannot tell you whether a result will matter to the people working on this problem.

This makes AI review best used at two moments: early, as a rigorous self-check before you trouble human readers, and late, as a final sweep for structural or clarity issues you have become blind to through familiarity. It is not a substitute for expert human judgment on the substance of the work — it is a way to arrive at the human reader with a stronger draft.

For a broader discussion of what AI tools currently do and do not do in the peer review process, see our overview of peer review and the options available for AI peer review.

Acting on feedback without losing your voice

Receiving feedback is disorienting in a specific way: every comment arrives with the implicit authority of a reader who found the paper hard to follow, disagreed with a claim, or wanted more evidence. It can feel like you are obliged to accept all of it. You are not.

A useful mental model: feedback is data about reader experience, not a prescription for how to fix the paper. A reader who says “this section confused me” has told you something real and worth taking seriously. A reader who says “cut section three” has told you what they would do if they were writing the paper. Those are different things.

Group comments before you act. Read all the feedback before changing anything. Often what looks like ten separate problems is three underlying issues appearing in different places. Acting comment by comment produces patchwork revisions; acting on root causes produces a better paper.

Distinguish the claim from the suggested fix. When a reader proposes a specific change, ask whether you agree with the underlying diagnosis first. If you do, you may find a better solution than the one they suggested. If you do not, the comment may still point toward something worth examining.

Write a response memo, even if you share it with no one. For each substantive piece of feedback, note briefly whether you are acting on it and why (or why not). This keeps you honest about motivated reasoning — “I’m not changing this because I’m right” is a different sentence from “I’m not changing this because I disagree with the premise that X, and here is why.” The second is a defensible position; the first is not.

Preserve what is working. Feedback under pressure tends to produce revision that fixes the critiqued parts while inadvertently weakening what was already strong. Before revising, note the sections your readers found clear, persuasive, or effective. Those are the anchors to protect.

Frequently asked questions

How early in the writing process should I share a draft?

As soon as the argument is clear enough to be critiqued, not before. For most papers, this means you have a complete draft — even a rough one — with a discernible structure and a stated claim. Sharing an outline for “big picture” feedback can work, but most readers find it hard to comment helpfully on something they cannot yet read as a whole.

How many rounds of feedback is normal?

For a journal article, two to three rounds of substantive feedback before submission is common — often one early round from a peer or collaborator, and one closer to submission from someone whose judgment you trust on the field. After that, feedback tends to be editorial rather than substantive. Revision after peer review adds further rounds.

What do I do if two readers give contradictory feedback?

Sit with both comments before acting on either. Often contradictory feedback points to genuine ambiguity in the paper — the passage or claim is unclear enough that readers fill it differently. The solution is usually to make your intent explicit, which resolves both comments at once.

How do I ask a senior colleague for feedback without seeming presumptuous?

Frame the request around a specific question rather than a general reading. “I’m working on a paper about X and I’m uncertain whether the framing in the introduction accurately represents the existing debate — would you have twenty minutes to look at those two pages?” is a request most people can say yes to. An open-ended “would you read my paper?” is harder, because the time commitment is unclear.

What should I do if I get feedback that feels wrong?

Take it seriously first. Readers who find something unclear or unconvincing are reporting their experience accurately, even if their suggested fix is off-base. Ask what would have helped them — the answer is often more useful than the original comment. If, after that, you still believe the passage is sound as written, note your reasoning and move on. Not every comment deserves a revision.

Can AI feedback replace peer review?

No — and any tool that claims otherwise is overselling. Peer review serves several functions: quality control, yes, but also positioning the work in the field, identifying the conversation it is entering, and assessing whether the contribution is significant. These require human expertise and community knowledge. What AI does well is the structural and clarity dimension of review: the questions a good peer reviewer would also ask before engaging with the substance. Think of it as a way to arrive at peer review with a stronger paper.

Getting feedback that moves the work forward

The researchers who get the most out of feedback tend to be the ones who are most intentional about it — they prepare before they share, they brief their readers clearly, and they treat the response as evidence rather than verdict.

Getting a range of perspectives matters. No single reader, however expert, sees every dimension of a paper. The advisor sees the methodology; the peer sees the argument from the outside; the AI reviewer applies a consistent frame to the whole structure; the editor notices what the writer has become too close to see.

PerfectPaper gives you referee-grade feedback in minutes — argument gaps, unsupported claims, clarity issues, missing context — the same questions a journal reviewer would ask, available before you submit. Use it to sharpen the draft before it reaches your human readers, so their time goes where it counts most.

Start a free review at PerfectPaper — no subscription required to begin.

You might also find these pages useful: research paper reviewer options compared and our guide to AI peer reviewers for academic work.