How to Write a PhD Statement of Purpose

Committees and supervisors read your statement of purpose (SOP) to answer three questions: Can you do research? Do you know what you want to work on? And why here? Most weak statements fail because they answer a different question, usually “what kind of person am I?”, with a childhood story and a list of adjectives.

A five-part structure

For a statement of about 800 words, this split works well. Adjust to the limit you are given.

1. The opening

Start with the problem, not with yourself. A reader should be able to guess your topic after two sentences. “I want to make medical image segmentation robust to differences between scanners” is far stronger than “I have always been passionate about artificial intelligence.”

2. Your research background

This is the heart of the statement. Choose two or three projects, ordered by relevance rather than chronology. For each one, answer four questions:

• What was the problem?
• What did you do? (Not “we”. Be clear about your own contribution.)
• What was the result? Give a number, an output, or a publication.
• What did it teach you, or what question did it leave open?

Compare these two sentences. Vague: “I worked on a machine learning project.” Specific: “I designed the evaluation protocol for our study and ran 120 experiments across four datasets, which showed that the improvement disappeared on out-of-distribution data.” The second tells a committee you can do research.

3. Your research direction

You are not committing to a fixed plan, but you must show you can think ahead. Describe a gap or open problem, an approach you might take, and which existing work it builds on. Name one or two papers or groups. Make sure the direction connects both to your past work and to the supervisor's work. A direction unrelated to either looks disconnected.

4. Why this program and this supervisor

Show that this is a considered choice, not a mass application. Name the supervisor, cite a specific paper of theirs, and explain how it connects to your direction. Mention a facility, dataset, or collaboration that lets you do something you could not do elsewhere. Praising the university's reputation or ranking adds nothing, because it applies to every applicant.

5. The closing

Two or three sentences: what you hope to achieve in the PhD, and where you want it to lead. Then stop. Do not summarise the whole statement again.

Mistakes that get statements ignored

• Clichés: “since I was a child”, “I have always been passionate”, quotations from famous people.
• Adjectives instead of evidence. “Hard-working and creative” proves nothing. A result does.
• A list of courses and skills with no evidence that you used them.
• Claiming credit for team results without saying what you personally did.
• A generic statement that could be pasted into any application. If a sentence could appear unchanged in another application, cut or customise it.
• Ignoring the word limit. Committees notice.

A note on using AI tools

AI can help you tighten or critique a draft. It should not write your story, invent achievements, or produce generic prose. Some programs restrict AI use in application materials, so check the rules and disclose if required. Committees read hundreds of statements, and generic, polished-but-empty text stands out.

Before you submit

1. Let it rest for a few days, then re-read it as a stranger would.
2. Check that every claim has evidence, a number, or a named output.
3. Read the first paragraph aloud. Could someone outside your field tell what you want to research?
4. Ask a supervisor or senior student to read it and comment.
5. Check spelling, grammar, and consistency (UK or US spelling, not a mixture).