How to write a research proposal that gets approved
A research proposal is the document you hand in before you are allowed to start. It contains no findings, so it is not judged on findings. It is judged on whether the question is worth asking, whether your plan can answer it, and whether the whole thing fits the weeks you actually have. Someone reads it and makes a decision. This guide is about that decision: who makes it, what published rubrics weight, how to argue a gap instead of asserting one, what makes a timeline believable, and what to do in the much more common case where the proposal comes back with changes.
Everything below is about the document itself. Once it is approved the work changes shape completely, and how to write a bachelor thesis picks up from there: the chapter structure, what a word limit costs when you go over it, how extensions and submission deadlines actually work, and the defense. One rule runs through both pages. Your own programme's requirements beat any general advice here, and if your department publishes a template or an assessment form, that document is the brief.
The decision at the end of it
Every proposal has the same reader profile underneath the local variations: someone who has to commit time or money to you before there is anything to show for it. On a bachelor or master's project that is a supervisor, sometimes with a committee behind them, deciding whether to take your topic and spend months on it. On a doctoral application it is a committee judging whether the project is doctoral in scope and whether you can carry it out. For a grant it is reviewers ranking your plan against other people's. None of them are reading for results, and all of them are reading for the same three things: a question they can restate, a method that could answer it, and evidence you know how long things take.
How formal that decision is varies more than most guides admit. At Vrije Universiteit Amsterdam the proposal is not a hurdle before a course, it is the course. Research Design carries 6 ECTS and is assessed as "A written Research Plan for the Research Project (70%, graded by thesis supervisor)" plus a "(Recorded) presentation of the Research Plan (30%, graded by the course tutor)", and the MSc thesis "starts immediately after completion of the Research Design course". Read the split again. The person marking seventy percent of your proposal is the person who will supervise you for the rest of the project. You are not persuading a stranger. You are showing your future supervisor how you think before anyone is checking your work.
The proposal is the cheapest place to be wrong. Changing the question in a three page plan costs an afternoon. Changing it after you have collected data costs a semester, and by then the plan has become the thing your work is measured against.
What a proposal is actually marked on
Most departments never publish their weights, so students guess, and the guess is usually that the idea carries the document. Justus Liebig University Giessen publishes a full rubric for MA thesis proposals, criterion by criterion, with the percentages attached. It is worth reading even if you study somewhere else, because it says out loud what most assessors are doing quietly.
| Criterion | Weight |
|---|---|
| Research overview: summarises pertinent research, precise references, covers findings and the methods behind them, names open questions | 10% |
| Experimental design and its suitability for investigating the research question | 10% |
| Methods of analysis specified in detail | 10% |
| Comprehensibility for a reader who did not attend the seminar | 10% |
| Linguistic accuracy and coherence | 10% |
| Precision and relevance of the research question | 5% |
| References; adequate literature research documented | 5% |
| Length | 5% |
| Title in conformance with conventions and self-explanatory | 5% |
| Relevant sections included | 5% |
| Organisation and structure | 5% |
| Formatting | 5% |
| Final version only: degree to which feedback has been taken into account | 15% |
Source: Assessment criteria for MA thesis proposals, Justus Liebig University Giessen. Three things fall out of it. The idea, listed there as "precision and relevance of research question", is worth five percent. Everything about whether a stranger can follow your plan, the design detail, the analysis detail, comprehensibility, language and structure, comes to forty. And the single heaviest line is not about the document at all: fifteen percent for what you did with the feedback. Its top band reads "completely (either incorporated or justification for not including suggestions provided)", which is the whole relationship in one parenthesis. You are allowed to disagree, in writing, with reasons. You are not allowed to ignore.
The same priorities hold further up the ladder. The doctoral committee at KU Leuven's Faculty of Law publishes its assessment criteria, and two are worth memorising. Originality is read as "the degree to which the proposed research goes beyond the state of the art in the area". Method is one flat sentence: "The proposed research method must be feasible within the proposed time frame." Not ambitious, not interesting. Feasible, inside the schedule you yourself proposed.
Those documents also settle the length question better than any general range. Giessen puts the top band at 1,500 to 2,000 words, calls 1,200 to 1,500 merely acceptable, and fails anything under 1,200. KU Leuven caps a doctoral proposal at "6000 words (including footnotes but excluding bibliography, and table of contents)". One more detail from Leuven is worth checking against your own rules before you shrug at your title: "The proposal's title is not a working title, but ought to cover the research contents", because approved proposal titles are published. The advice that your title is provisional is true in a lot of programmes and false in some. Find out which one you are in.
The gap has to be argued, not asserted
"Little research has examined X" is the most common sentence in student proposals and the easiest one to disprove. A reader who knows the field runs a single search, finds three papers, and now doubts everything else you have claimed to have read. A gap is a claim about a body of literature, and a claim needs evidence.
Leuven's committee makes the point precisely: "it is not the exhaustiveness of the suggested sources which is most relevant, but rather the degree to which the relevant sources are truly incorporated into the research proposal". A long reference list is not an argument. Four moves make one. Say what is established, naming the papers that agree. Say what is contested, naming the papers that disagree and the reason they disagree. Say what none of them do, precisely enough that a reader could check. Then say why that missing piece is worth having now, whether because new data exists, because a population nobody sampled has become reachable, or because the disagreement cannot be settled without it.
The shape, with placeholder findings. "Cargo bike delivery has been shown to cut last leg emissions in dense city centres, and the three studies below agree on the direction of the effect. They disagree about cost, because two measure fleet operating cost and one measures cost per parcel. None of them study mid-sized cities, where depots sit further out and the final leg is longer, so the assumption that the dense-centre result transfers has not been tested. Testing that transfer is what this project does." Every sentence there would point at a specific paper you have read. Written with sources you have not opened, it is the exact paragraph that gets a proposal returned.
Feasibility is the part they read hardest
Feasibility is where ambition meets the calendar, and it is the one part of a proposal that writing well cannot rescue. Four questions decide it, and a reader is asking all four whether or not your template has a section for them.
- Access. Do you already have the data, the participants, the archive, the permission? Something you are confident you can get is not the same as something you have.
- Skill. Can you run the analysis you are proposing, today, or is there a learning period hiding in your schedule that the schedule does not show?
- Calendar. Does collection finish early enough that analysis and writing have room, including the approvals, review windows and other people's holidays that sit between you and your own deadline?
- Dependency. What is the single thing this project cannot survive losing, and what happens if it does not arrive?
Answer the last one in the document. A plan with a stated fallback reads as thought through; a plan with no fallback reads as untested. Then describe what you will collect, from whom, how, and how you will analyse it, in enough detail that a reader can judge the fit rather than take your word for it. The components a reader expects to see, and a checklist for covering them, are in how to write a methodology. If your question itself is still too wide for any of this to be answerable, fix that first in how to write a research question.
A quick feasibility test: write the project as one sentence in the form "I want to find out whether X affects Y, in this population, using this method, within these weeks." If you cannot fill every slot, the proposal is not concrete enough to approve yet.
A timeline that reads as real
The timeline is the section experienced readers scan first and students write last. It is the only part of the document that cannot be bluffed by good prose, because it is arithmetic against a date everybody in the room knows.
Build it backwards from the submission deadline, and make four things true of it. Name deliverables, not activities: "literature map, twenty papers grouped by theme" is checkable in a way that "reading" never is. Put the dependency first, so the thing that could sink the project fails early enough to be survivable. Leave the last stretch clear, because a schedule with no slack at the end tells a reader you have not done this before. And show other people's time as time: an ethics review, a supervisor's response window, a data request that a stranger has to approve.
A credible one term schedule, built backwards from the deadline. Weeks 1 to 3, literature mapped and the question locked. Weeks 4 to 7, collection or the build, with the risky dependency resolved in week 4 rather than week 7. Weeks 8 to 10, analysis. Weeks 11 to 13, full draft in the supervisor's hands. Week 14, revision, references and formatting, held clear on purpose.
When it comes back with changes
Almost nobody tells students this, so it lands as a rejection: coming back with changes is the normal path. The document is a plan, the reader's job is to stress test it, and a plan nobody pushed on has usually not been read closely.
Where the proposal is a graded course, the second round is a formal event with its own rules. Vrije Universiteit Amsterdam's Research Design course resits "the presentation and a revised version of the research proposal as well as the Academic Writing component", and it says the quiet part out loud: "expectations of tutors and supervisors may increase given the extra time available". You are not resubmitting the same standard. You are resubmitting against a higher one.
So treat round two as the graded event, because on Giessen's rubric it literally is the heaviest line on the page. Three habits carry it. Answer every comment explicitly, in a short response note that says what you changed and where, so nobody has to hunt for your edits. Where you did not take a suggestion, say so in that note with the reason, which is the behaviour the rubric's top band describes rather than a risk you are running. And never leave a comment silently unaddressed, because a revision that skips one is the version that comes back a third time.
Where we would spend the last hour
Here is our opinion, and it goes against most guides on this subject. If you had one hour left before submitting, we would not spend it on the literature review. We would spend it on the timeline and on one honest paragraph about what could go wrong with your access to data. The reason is in the rubrics above: the only student-facing weights we could find published put the research question at five percent, put design, analysis and clarity at forty, and put the handling of feedback at fifteen. A supervisor approving a proposal is not buying an idea, they are agreeing to spend months of their own time on your project, and everything they are actually scored on when they read you is about whether that time will be well spent. The literature section can be strengthened in week two of the project. A timeline that does not fit the calendar cannot, because by the time it fails, the calendar has already run out.
Real literature behind the gap
The slow, error-prone part of a proposal is the same everywhere: finding real literature for the gap argument, reading enough of it to say what each paper stops short of, and keeping the references honest. That is what CiteOwl is built for, on a verify-first principle. It searches actual literature and reads what it finds, so the papers behind your citations are real rather than generated from memory, and every claim it writes links to a source it actually read, with the supporting quote shown.
It will help you surface and cite the work your gap argument leans on, and draft the sections around it, the problem statement, the objectives, the method sketch, which you review one change at a time as a plain diff and accept or reject. The line it does not cross is the thinking. The question, the method, and the judgement about what is missing from a field stay yours, which is just as well, since those are the parts your supervisor is about to decide on.