For GRFP applicants before they submit
NSF GRFP Mock Review of Your Application
Drop your personal statement and research plan. Get a reviewer's read on both criteria, your lines quoted, and the five fixes that matter most.
Free to start. No card needed.
Research Plan · Intellectual Merit
We developed a pipeline to classify coral bleaching from drone images, and the results were promising.
In graduate school I will extend this work to new reef sites.
- Whose work?"We developed" hides your part. Say what you built.
- Promising how?Give a result a reviewer can weigh: a number or a finding.
What you'll get
A reviewer's read of both statements before you submit
- Intellectual Merit and Broader Impacts read like a reviewer, with your lines quoted
- A skim test: what a busy reviewer takes from your first sentences alone
- Your five most important fixes, in order, across both statements
Before and after
Your statements in. A reviewer's notes back.
A three-page personal statement and a two-page research plan as PDFs. A chemistry senior proposes a cheaper catalyst for breaking down plastic waste, after two summers of lab research and three years running a tutoring program.
Intellectual Merit · River's estimate: Very Good Strength: two summers of lab work and a poster at a regional meeting. Weakness: "we optimized the reaction" never says which step was yours. Broader Impacts · River's estimate: Good The tutoring program is concrete. The plan's outreach is one sentence, with no partner and no numbers. Skim test: the plan's first sentences never state the hypothesis. Fix first: state the hypothesis in the plan's opening paragraph.
Ratings are River's estimate from your text, not a prediction of NSF's decision. Every note quotes your words.
Why it works
How do GRFP reviewers read an application?
A GRFP application review is the read a panel reviewer gives your two statements. River gives that read to your Personal, Relevant Background, and Future Goals Statement and your Graduate Research Plan Statement before you submit. It weighs both against NSF's two criteria, Intellectual Merit and Broader Impacts, quotes the lines that help and the lines that hurt, and ranks your five most important fixes. Your files stay exactly as you wrote them, and you decide what changes.
Written feedback from NSF has disappeared. For the 2026 competition, applicants received a band rank and ratings for each criterion, and NSF says "No written reviewer comments were collected" (NSF). Both criteria count: NSF's solicitation says each "is necessary but neither, by itself, is sufficient," so a strong lab record cannot carry thin Broader Impacts. River checks your evidence against the factors NSF lists, such as planning research, working independently and taking initiative, and flags every claim that no sentence proves.
Run it a week or more before your field's deadline, while there is still time to act on the notes. If Broader Impacts comes back weak, the Broader Impacts builder turns the activities you already do into specific plans. A thin or vague plan can be rebuilt with the research proposal writer. For a second human read, give a friend or mentor the GRFP review rubric, which scores a draft against the same criteria.
What lands in your doc
What's in your GRFP review
- Intellectual Merit strengths and weaknesses, each tied to an NSF factor and a quote
- Broader Impacts read the same way, past activities and future plans checked for specifics
- River's estimated rating for each criterion, with one plain sentence on why
- A skim test listing your opening sentences, and what a reviewer would miss
- Where your two statements disagree, and where the plan ignores your experience
- Your five most important fixes, ranked, each naming the statement and paragraph
How it works
From draft to reviewer's notes
Drop your statements
Drop the personal statement and research plan as PDF or Word files, or paste the text.
River reads both
It reads every paragraph before writing, noting your role, your results and each plan.
Get your review
The review lands in your doc: both criteria, the skim test, and your ranked fixes.
Fix and recheck
Make the changes you agree with, then ask River to rewrite one paragraph or cut to length.
How to review a GRFP application yourself
Start with the skim. NSF tells reviewers to expect about 20 applications at first and three to five more later (NSF), so each one gets a fast first read. Copy the first sentence of every paragraph into a list, along with any bold text, and read only that. A reviewer should learn your question, your hypothesis, what you have already done and what you will do for others. If they would not, move those sentences up.
Then read for evidence, one NSF factor at a time. For Intellectual Merit, mark each claim of skill with the sentence that proves it: a method you ran, a result with a number, a talk or a paper. Circle every "we" and ask whether a reviewer can tell your part from your lab's. NSF reminds reviewers that applications "are not research proposals, but fellowship applications," so the plan shows how you think, and it is not a contract.
Last, test Broader Impacts the way you tested your research. Past activities need numbers and your own role: how many students, for how long, and what changed. Future plans need a who, a where and a how often, and they should grow out of what you have already done. Check that both statements use the exact headings "Intellectual Merit" and "Broader Impacts", and that the plan and the personal statement tell the same story about you.
Questions GRFP applicants ask
Common questions
What do GRFP reviewers look for?
Evidence for two criteria. Intellectual Merit is your potential to advance knowledge: planning and doing research, working independently and in a team, communicating results, initiative, creativity and persistence. Broader Impacts is your potential to benefit society, such as mentoring, outreach and teaching. NSF says each criterion is necessary and neither is enough alone (NSF 26-526).
Does NSF give feedback on GRFP applications?
Less than it used to. For the 2026 competition, every applicant received a band rank and ratings for Intellectual Merit and Broader Impacts, and NSF says no written reviewer comments were collected. A review before you submit is written feedback you can act on. If you have last year's ratings, the ratings decoder explains them.
Is the rating River gives the same as NSF's?
No. River's rating is an estimate from your text alone, using the words NSF's reviewers have long used, from Excellent down to Poor, to show which criterion needs the most work. Real panels also read your letters, transcripts and record, and compare you with other applicants in your field. Treat it as a pointer to your weakest section, not a prediction.
Can River review just one statement?
Yes. Drop only the personal statement or only the research plan and River reviews that one, says the other is missing, and skips the check that the two agree. Reviewing both together is more useful, because reviewers read them as one application. Weak spots often show up as a mismatch, such as a plan that never draws on the experience you describe.
Is it OK to have AI review my GRFP statements?
NSF's application rules do not forbid it, and you certify that both statements are your own original work. A review keeps that true: River quotes your lines and suggests changes, and your files stay exactly as you wrote them. NSF's notice on generative AI says applicants are responsible for the accuracy of anything AI helped with.
Are my statements used to train AI?
No. Under River's privacy policy, nothing you write or upload is used to train AI models, and River's agreements with its AI providers bar them from training on it too. Your statements and the review stay in your private workspace, where you can edit or delete them at any time.
Your next step
More for your GRFP application
Find the weak paragraph before a reviewer does
Drop your statements. Every note quotes your words and says what to change, while there is still time to change it.