How to Write a Research Proposal (Section by Section)

A research proposal is a plan a stranger can judge. For a PhD scholarship that stranger is an admissions panel. For a seed grant it is a reviewer who may not work in your exact subfield. Both are asking the same three questions: is this problem worth solving, can this person actually do it, and is the plan specific enough to fail in an interesting way if the idea is wrong?

Write to the page limit you were given

A PhD research statement is often two to four pages. A small internal grant might allow five to eight. A national scheme will give you a template with strict headings and word caps. Use their headings when they exist. The outline below is for the common case where you are told the length and little else. Scale the sections to fit. Do not add a section the funder did not ask for if you are already over the limit.

Problem, then aims

Open with the problem in language a neighbouring specialist can follow. Then state the aims as outcomes, not activities. “Explore deep learning for medical images” is a topic. “Test whether a model trained on one hospital’s scans keeps its accuracy on a second hospital, and measure how much local fine-tuning that takes” is an aim a reviewer can score.

Two or three aims is enough. Each aim should imply a comparison, a material or dataset, and a way to know whether it worked. If you cannot say what “worked” means, the aim is not ready.

What is already known, in one tight passage

This is not the place for a survey. Name the line of work your project sits next to, say what that work has established, and state the hole in one or two sentences. The hole has to be one you can actually work on. A useful way to draft it is: “Recent studies show A. They do not establish B, because C. This project tests B under conditions D.”

If you have not yet done the search that justifies “they do not establish B”, stop and do that first. A proposal that claims a gap from memory is the most common reason a careful reviewer loses confidence. The companion guide How to find a research gap is the check to run before you write this paragraph.

A method a reviewer can believe

Describe the work in the order you will do it, aim by aim. For each aim, name the data or materials, the comparison, and the criterion. A sketch is more convincing than a list of techniques.

• Name the baseline. A proposal that only describes the new method gives the reviewer nothing to compare it with.
• Say where the data comes from. “Public datasets” is weaker than the specific collection, and weaker still if you need approval you have not started.
• Separate what you will build from what you will measure. Reviewers fund the measurement.
• One paragraph on ethics or access if the work uses people, clinical records, or a facility you do not already have.

Timeline, risks, and why you

A short timeline is a feasibility argument. Put the risky aim early enough that a failure still leaves a result. For each serious risk, write the fallback in the same sentence: what you will do instead, not that you “will monitor risks”. Then, briefly, why this team or this applicant can do the work: the relevant skill, the data already in hand, the supervisor or collaborator who covers the part you do not. A PhD proposal can point at coursework and a small pilot. A grant proposal should point at prior results, not at enthusiasm.

Significance without the inflation

Say who can use the result and what they could do that they cannot do now. Skip “transformative”, “novel paradigm”, and “first ever” unless you can defend the word with the search you already did. A modest, true claim scores higher than a large one the reviewer does not believe.

Before you submit

1. Read it as the reviewer. Can you score significance, feasibility, and clarity from this draft alone, without asking the author a question?
2. Check the administrative rules. Eligibility, budget categories, required headings, and the page limit. Strong science in the wrong template is still a desk reject.
3. Cut anything that does not serve an aim. Background that does not set up the hole should go.
4. Ask one person outside your group to mark the sentence where they got lost. Fix that sentence, not the paragraph around it.