Writing for the Reviewer

A grant reviewer is a subject-matter peer, usually a volunteer, reading a stack of applications against published criteria on a deadline. Most spend hours rather than days on any single proposal. Write so that a tired, competent, non-specialist reader can find every criterion without hunting for it.

Current figures — verified 2026-08-11

Item	Value	Source
Average NSF proposals submitted per award	2.3 proposals per award	NSF, Overview of the Proposal and Award Process
New NSF principal investigators funded on first attempt	36% in 2020	NSF, Overview of the Proposal and Award Process

These figures change. Verify against the linked source before relying on them. Report an outdated figure

Key takeaways

• Grant reviewers are mostly untrained volunteers reading against a published rubric.
• One or two assigned reviewers argue your case to a room that skimmed.
• Overall merit is a judgment, not the average of criterion scores.
• Agreement between reviewers on strong proposals is close to zero.
• Control what is controllable: structure, labeling, and plain sentences.

Who are grant reviewers, and how do they get assigned?

Grant reviewers are external subject-matter peers recruited by the funder, usually serving without pay or for a modest honorarium alongside a full-time job. Within writing the proposal, the reviewer is the actual audience — not the program officer, not the agency, and not the person who forwarded you the announcement.

Reviewer preparation is thinner than applicants assume. In a survey of 258 grant reviewers working in 22 countries, 85 percent had never been trained in grant review, only 7 percent were given protected time for it, 74 percent received no academic recognition, and just 16 percent said guidance from funders was very clear (Schroter, Groves & Højgaard, BMC Medicine 8:62, 2010).

Assignment structures differ but converge on one shape. The National Science Foundation typically identifies at least three external reviewers per proposal, working as ad hoc reviewers, a panel, or both (NSF, Overview of the Proposal and Award Process). The National Institutes of Health assigns at least three reviewers who write critiques and preliminary scores before the meeting, after which the wider study section votes (NINDS, Scoring and Summary Statements).

The University of Michigan’s proposal guide states the reader model without flattery: “It is well to assume that your reader is a busy, impatient, skeptical person who has no reason to give your proposal special consideration and who is faced with many more requests than he can grant, or even read thoroughly” (University of Michigan ORSP, The Proposal Writer’s Guide).

How much time does a grant reviewer spend reading?

A grant reviewer spends hours on a proposal, not days, and the total includes reading, note-taking, writing the critique, completing score sheets, and attending the panel. Asked how long the whole process took for a single biomedical proposal, 15 percent of reviewers reported three hours or less and 30 percent reported four to six hours (Schroter et al., BMC Medicine, 2010).

Panel members who were not assigned your proposal spend far less. Their scores nonetheless count toward the final average, and survey evidence identifies “limited participation from unassigned reviewers” as a recurring weakness of panel discussion (Gallo et al., Research Integrity and Peer Review 5:7, 2020). A vote cast on the strength of an abstract and someone else’s critique is still a vote.

Length works against a grant reviewer’s accuracy. In an evaluation of one Norwegian health research funder, moving from long single-stage proposals to short first-stage proposals produced scores with greater reliability and agreement, and cut average reviewer time per applicant by 28 percent (Seeber, Svege & Hesselberg, Research Evaluation, 2024). Extra pages are not free; they are spent out of the same attention budget that your strongest argument needs.

Why does one grant reviewer argue your case to the whole panel?

In panel review, the assigned grant reviewer is your proxy in a room where most members have not read you closely. The National Institutes of Health tells applicants this directly: “They summarize your application and guide the review panel’s discussion” (NIH, General Grant Writing Tips). The same guidance adds the reason it matters: “All reviewers are important to you because each gets one vote.”

The practical consequence is a distinct writing objective. A proposal must be summarizable — it must hand the assigned reviewer a defensible one-paragraph case, plus two or three concrete strengths they can name aloud when challenged. Clarity is necessary but not sufficient; the case has to be portable.

Discussion moves fewer proposals than applicants hope. Studies estimate that a proposal’s funding status shifts from before to after panel discussion for only 10 to 13 percent of proposals (Gallo et al., 2020). What does move scores is talk in which reviewers reference each other’s scoring habits — the canonical example being “your comments are meaner than your score.” A written critique that a reviewer cannot reconcile with a generous number is where advocacy breaks down.

Order matters too, and not in your favor. In an innovation fund competition where evaluators were quasi-randomly assigned and could not influence each other, a proposal evaluated first “requires total assets in the top 10th percentile to merely equal the evaluation of a proposal in the bottom 10th percentile that is not evaluated first” (Bian, Greenberg, Li & Wang, Management Science 68(1), 2022).

How do grant reviewers turn criteria into a score?

Grant reviewers score each published criterion separately, then form an overall judgment that is explicitly not the average of those criterion scores. The National Institutes of Health states the rule plainly: “The impact score should reflect the reviewer’s overall evaluation, not a numerical average of individual criterion scores” (NIH, Scoring System and Procedure).

Two implications follow. First, a proposal “does not need to be strong in all categories to be judged likely to have major impact” — the same NIH guidance offers the example of a project that is not innovative but is essential to advance a field. Second, the overall judgment decomposes: likelihood of success is drawn mainly from the investigator, approach, and environment criteria, while durable influence is drawn mainly from significance and innovation (NIH, Overall Impact versus Significance).

Adjectival systems behave the same way. NSF states that “a proposal doesn’t need to receive all ‘excellent’ scores to be funded. Conversely, receiving all ‘excellent’ scores is no guarantee of funding” (NSF, Overview of the Proposal and Award Process).

Most federal discretionary announcements publish point-weighted criteria instead. Weight is the instruction: allocate narrative space in rough proportion to published points, which is a different discipline from writing whatever you know most about. Where the points live is set out in the announcement itself, which is why reading a NOFO backward from the review criteria is the standard move.

How much do grant reviewers agree with each other?

Grant reviewers evaluating the same strong proposals agree with each other about as much as chance would predict. Replicating the full NIH process with 43 oncology reviewers, 25 previously submitted R01 applications, and 83 ratings, researchers found an intraclass correlation of 0 for preliminary ratings and a Krippendorff’s alpha of 0.024, against a conventional acceptability threshold of 0.7 (Pier et al., PNAS 115(12), 2018).

The authors state the finding in ordinary language: “two randomly selected ratings for the same application were on average just as similar to each other as two randomly selected ratings for different applications.” Their reviewers also “rated unfunded applications just as positively as funded applications.”

State the limit of that finding honestly. The sample contained only high-quality applications, so the study shows that reviewers do not reliably distinguish good from excellent — not that they cannot distinguish sound work from seriously flawed work. Randomness at the top of the pile is a different claim from randomness everywhere.

Funder-side analysis points the same way. A retrospective study of one Australian funding round found that 59 percent of 620 funded grants were sometimes not funded once random variation in panel members’ scores was taken into account, with 29 percent of all proposals landing in a “sometimes funded” band (Graves, Barnett & Clarke, BMJ 343:d4797, 2011). One score is a noisy measurement, which is the central fact behind what to do when a grant is declined.

What writing choices match how grant reviewers read?

Writing choices that match grant reviewer behavior are structural rather than stylistic: mirror the published criteria, label everything, front-load every unit of text, and define terms where the reader is, not where you first used them. The National Institutes of Health puts the underlying principle first: “Reviewers are accustomed to finding information in specific sections… This creates an efficient evaluation process and saves reviewers from hunting” (NIH, General Grant Writing Tips).

The table below maps documented grant reviewer behavior to the writing tactic that answers it.

Reviewer behavior	Documented basis	Writing tactic
Scores criteria one at a time	Criterion scoring is separate from overall judgment	Headings that repeat the criteria, in the published order
Skims and jumps between sections	Reviewers move around rather than read start to finish	Spell out each acronym on first use within every section
Reads under time pressure	Most spend under six hours per proposal	Topic sentence first; sentences of about twenty words
Votes without full reading	Unassigned panelists still score	An abstract that carries the whole argument on its own
Summarizes your case aloud	Assigned reviewers guide the discussion	A quotable one-paragraph statement of what and why
Loses accuracy as length grows	Shorter proposals scored more reliably	Cut to the point limit, not to the page limit

Two mechanics deserve emphasis. Acronyms should be spelled out on first use in each narrative section, because reviewers do not read linearly. And headings should reuse the funder’s own wording — NSF has returned proposals without review for omitting literally labeled sections, requiring that a Project Summary carry separate sections labeled “Overview,” “Intellectual Merit,” and “Broader Impacts,” and that the Project Description contain a heading for “Broader Impacts” (NSF, Dear Colleague Letter 15-095).

Register should match the section, not a single readability target. Plain language belongs in the title, summary, and public-relevance statement; the technical narrative is written for a scientifically literate reader. The abstract and project summary is where that split is most visible, because it is the one page every panel member reads.

What goes wrong when a proposal ignores the reviewer?

Proposals ignore the grant reviewer in six recurring ways, and each one converts reviewer effort into reviewer irritation. None of these are argument failures — they are handling failures that degrade an argument the writer already had.

• Criteria buried. Content that answers a scored criterion sits under a heading the reviewer does not associate with that criterion, so the criterion reads as unaddressed.
• Space allocated by enthusiasm. The section the writer enjoys gets ten pages and eight points; the section worth thirty-five points gets three pages.
• Undefined shorthand. Program names, models, and acronyms defined once on page four and used unexplained on page nineteen.
• Sloppiness read as a proxy. NIH warns that “if the application is sloppy or disorganized, reviewers may think your research could be conducted in the same manner” (NIH, General Grant Writing Tips).
• No summarizable claim. Nothing in the document gives the assigned reviewer a sentence to say out loud, so the case reaching the panel is whatever they improvise.
• Length as insurance. Extra pages added to cover every objection, spending attention that the strongest argument needed.

The profession’s own scoring of grant writing puts these in order. The Grant Professional Certified writing exam weights persuasive argument at 34 percent, organizing ideas at 22 percent, and conveying ideas clearly at 18 percent, against 10 percent for standard written English and 4 percent for following formatting guidelines (Grant Professional Certification Institute, Competencies Tested). Formatting is the smallest weight in the rubric and the largest cause of unforced loss, because a formatting failure is caught before scoring begins — the subject of building a compliance matrix.

Frequently asked questions

Do grant reviewers read the entire proposal?

Assigned grant reviewers read the entire proposal, usually twice — once for orientation and once in detail. Other panel members frequently do not, and survey evidence describes limited participation by unassigned reviewers who nonetheless score. Write the abstract and the criterion headings so that a partial read still produces a defensible score.

Are grant reviewers paid?

Grant reviewers are typically unpaid or paid a modest honorarium. Survey data found that fewer than half of reviewers’ institutions encouraged the work, only 7 percent received protected time, and 74 percent got no academic recognition. Reviewing is discretionary effort layered on top of a full workload, which is the practical reason clarity is rewarded.

Should you copy the funder’s exact headings?

Copy the funder’s exact headings whenever the announcement supplies them. NSF has returned proposals without review for missing literally labeled sections, and reusing the funder’s vocabulary removes the translation step between what you wrote and what the reviewer is scoring. Mirroring headings is a navigation aid, not flattery.

Does a low score mean the proposal was weak?

A low score means the proposal was scored low by the specific reviewers who read it. Replication research found no reliable agreement among reviewers evaluating identical strong applications, and one national analysis found most funded proposals sat in a band where outcomes shifted with panel composition. Read the written critiques for signal; treat the number as noisy.

How long should a grant proposal be?

A grant proposal should be as long as the point allocation justifies and no longer than the page limit permits. Evidence from one funder’s evaluation redesign found that shorter proposals produced more reliable reviewer scores. Pages spent on low-weight material draw attention away from the criteria that decide the outcome.

Do grant reviewers know your field?

Some grant reviewers know your exact subspecialty and most do not. NIH instructs applicants to write for two audiences at once: the majority who are not experts in your niche and the smaller number who are. Terms of art get defined, and the significance of the work gets stated rather than assumed.

Related topics

• Writing the Proposal — the hub covering each section of a proposal and what it must prove
• The Anatomy of a Grant Proposal
• Writing the Project Design
• How Funders Decide Who Gets Money
• Research Grants at NIH and NSF

Sources

1. Sara Schroter, Trish Groves and Liselotte Højgaard, “Surveys of current status in biomedical science grant review: funding organisations’ and grant reviewers’ perspectives,” BMC Medicine 8:62, October 20, 2010. https://bmcmedicine.biomedcentral.com/articles/10.1186/1741-7015-8-62 (accessed 2026-08-11)
2. Elizabeth L. Pier, Markus Brauer, Amarette Filut, Anna Kaatz, Joshua Raclaw, Mitchell J. Nathan, Cecilia E. Ford and Molly Carnes, “Low agreement among reviewers evaluating the same NIH grant applications,” PNAS 115(12):2952–2957, March 5, 2018. https://www.pnas.org/doi/10.1073/pnas.1714379115 (accessed 2026-08-11)
3. Nicholas Graves, Adrian G. Barnett and Philip Clarke, “Funding grant proposals for scientific research: retrospective analysis of scores by members of grant review panel,” BMJ 343:d4797, September 27, 2011. https://www.bmj.com/content/343/bmj.d4797 (accessed 2026-08-11)
4. Stephen A. Gallo et al., “Grant reviewer perceptions of the quality, effectiveness, and influence of panel discussion,” Research Integrity and Peer Review 5:7, May 15, 2020. https://link.springer.com/article/10.1186/s41073-020-00093-0 (accessed 2026-08-11)
5. Marco Seeber, Ida Svege and Jan Ole Hesselberg, “Improving the efficiency of research proposals evaluation: A two-stage procedure,” Research Evaluation, May 15, 2024. https://doi.org/10.1093/reseval/rvae020 (accessed 2026-08-11)
6. Jiang Bian, Jason Greenberg, Jizhen Li and Yanbo Wang, “Good to Go First? Position Effects in Expert Evaluation of Early-Stage Ventures,” Management Science 68(1):300–315, January 2022. https://ideas.repec.org/a/inm/ormnsc/v68y2022i1p300-315.html (accessed 2026-08-11)
7. National Institutes of Health, General Grant Writing Tips. https://grants.nih.gov/grants-process/write-application/general-grant-writing-tips (accessed 2026-08-11)
8. National Institutes of Health, Scoring System and Procedure. https://grants.nih.gov/grants/peer/guidelines_general/scoring_system_and_procedure.pdf (accessed 2026-08-11)
9. National Institutes of Health, Overall Impact versus Significance. https://grants.nih.gov/grants/peer/guidelines_general/impact_significance.pdf (accessed 2026-08-11)
10. National Institute of Neurological Disorders and Stroke, Scoring and Summary Statements. https://www.ninds.nih.gov/funding/after-you-submit/understanding-review-process/scoring-and-summary-statements (accessed 2026-08-11)
11. U.S. National Science Foundation, Overview of the Proposal and Award Process. https://www.nsf.gov/funding/overview (accessed 2026-08-11)
12. U.S. National Science Foundation, Dear Colleague Letter 15-095, Communication About Automated Compliance Checking and Proposal Submission. https://www.nsf.gov/funding/information/dcl-communication-about-automated-compliance-checking-proposal/nsf15-095 (accessed 2026-08-11)
13. University of Michigan Office of Research and Sponsored Projects, The Proposal Writer’s Guide. https://sites.lsa.umich.edu/nextprof-science/wp-content/uploads/sites/130/2018/04/3-University-of-Michigan-Proposal-Writers-Guide.pdf (accessed 2026-08-11)
14. Grant Professional Certification Institute, Competencies Tested. https://www.grantcredential.org/the-exam/competencies-tested/ (accessed 2026-08-11)