CiteOwl

Blog/Thesis and dissertation

How to write a bachelor thesis, step by step

The CiteOwl Team· Updated 20 September 2026· 16 min read

You write a bachelor thesis the way you write a research paper, stretched over a term, plus four things a paper never has. A proposal somebody has to approve. A supervisor who marks how you worked, not only what you wrote. A chapter structure your department fixes, and a defense at the end. The writing method does not change. The planning does, so start from your faculty's handbook and work backwards from the submission date.

Pick a question your department will approve

A thesis topic has to clear two bars at once. It has to be narrow enough to finish in the weeks you have, and it has to have enough published work behind it that you are not starting from nothing. "Climate change and the economy" is not a topic, it is a library. "The effect of carbon pricing on small manufacturers in one country between 2015 and 2020" is something you can finish in a term.

The narrowing itself works the way it does for any paper, and the fastest check is a two-minute literature dipstick: search your question's exact phrases in a free index and read the count, because a near-empty result usually means you used your words instead of the field's. That test, with a worked narrowing through four drafts, is step 2 of the research paper method. What is different at thesis scale is the other bar. You are committing a term, so the question has to survive the access and the skills you actually have.

If your method needs interviews with 200 executives, or a dataset behind a licence your university does not hold, the topic is fighting you, and your supervisor will say so in the first meeting. It is cheaper to hear it from yourself in week one.

The proposal, and what approval buys

Most programs ask for a short proposal before the thesis proper, and in most European and international programs it is how the department formally signs off on your topic. You cannot start until someone with the authority has agreed to it. In German-speaking and some other European universities the same document is called an Exposé; in English-language programs it is a thesis proposal or a research proposal, and the conventions carry across.

It is a separate document with its own rules, its own assessment and its own failure modes, so it has a guide of its own. How to write a research proposal covers what published assessment rubrics weight, how to argue a gap instead of asserting one, what makes a timeline believable, and what to do in the very common case where it comes back with changes. Read that before you write a word of it, then come back here for what happens after approval.

Two things are worth knowing before you go. The revision round is normal, and a proposal that comes back with comments has been read properly, which is more than some get. And approval is the last moment when scope is cheap: after it, every widening of the question costs you reading you have already done.

Your supervisor is marking how you worked

Treat your supervisor as a collaborator, not an examiner you avoid until the end. They have read dozens of theses and can tell in five minutes whether a topic is too broad or a method is unrealistic. Come to meetings with specific questions and a short written update, not "how's it going", and send the document rather than describing it. Agree early on how often you will meet and what they expect to see each time.

That relationship is part of the assessment in more programs than students realise, because a supervisor is judging how far you drove the work as well as how the finished document reads. It is the reason the honest answer to "can I just have AI write this" is the one in can AI write my thesis: a thesis is examined as a process, not only as a file. A supervisor who has signed off on your question and your plan is also the best insurance you have against rewriting half the thesis in the final month.

The chapters, and what each one is for

Once your question is set, the document has a shape, and knowing it ahead of time turns the blank page into a set of labelled boxes to fill. The structure below is common across European and international bachelor programs, and it is not a law. Your department's template decides the required chapters, their order and their length, so confirm it before you build your document.

The percentages are a rough guide, not a quota. The rhythm is the point: you set up a question, show what is known, explain what you did, report what happened, then interpret it. Some fields merge results and discussion into one chapter, and a theoretical thesis may have no results chapter at all.

Two of those chapters lose the most marks, and both have guides of their own. The literature review is the one students write source by source, and "Smith found X, then Jones found Y, then Lee found Z" is a summary wearing a review's clothes; the synthesis markers reward, theme by theme, is in how to write a literature review with AI, and the screening pass that stops a chapter being built on abstracts is step 4 of the research paper method. The methodology chapter has to match your question rather than impress with it: a question about how widespread something is wants a survey, an experiment or a dataset, and a question about how people experience something wants interviews or cases. Its components are in how to write a methodology. Name your limitations there or in the discussion, because examiners trust a thesis that knows its own weaknesses far more than one pretending to have none.

How long, and how many sources

These are the two questions every student asks, and the honest answer to both is the same: check your department guidelines, because they vary by country, university and field, and no number from a blog overrides your faculty's handbook. Some typical ranges help you orient.

For length, many bachelor theses land somewhere around 30 to 60 pages, or roughly 8,000 to 12,000 words, though a quantitative thesis in a technical field is often deliberately shorter and a humanities thesis can run longer. A master's thesis is generally a step up, around 60 to 100 pages or 15,000 to 30,000 words, and the extra length is not padding: it is the room a deeper literature review, a developed framework and a fuller account of the contribution need. Faculties name figures far more specific than any range online. The University of Auckland, for instance, sets a 120-point master's thesis at 35,000 to 40,000 words. Your handbook usually fixes the word count, the page range, the margins and the font, and it is the only count that matters.

Inside that total, the split between chapters is the number worth writing down before you draft. In 58 published papers we measured, the introduction took a median 14 percent of the body and the methods 26, with the conclusion at under 4; the table of budgets at three document lengths is in step 5 of the research paper method. Scale it to your own word count, write the number beside each chapter heading, and the vaguest question in writing turns into arithmetic.

For sources, many bachelor theses cite somewhere around 20 to 40 references, weighted toward peer-reviewed papers and recent work, though a narrow technical question might need fewer strong sources and a broad review many more. The count is the wrong thing to optimise. A focused thesis built on sources you have read beats a padded list of papers you skimmed once. Ask your supervisor what they expect, and let your research question decide the rest.

What changes at master's level

If you are heading toward a master's thesis, you do not need to relearn the shape. The chapter list is nearly identical, with two common additions: a formal table of contents with lists of figures and tables, and a theoretical framework chapter that names the concepts, models or theory you analyse through. Some programs keep the framework as its own chapter at roughly a tenth of the body; others fold it into the end of the literature review. Either way, it is where you stop describing the field and start telling the reader how you will make sense of what you find. The literature review also grows, to roughly 25 to 30 percent of the body, and the abstract stretches to around 200 to 350 words.

What really changes is the bar inside each box. Four differences matter most.

• More depth, not more description. A master's thesis is usually longer, and the extra space goes into deeper analysis. Each chapter has to earn its pages.
• A bigger, more critical literature review. You are expected to command the field, not sample it, and to be critical about where the existing work falls short.
• An original contribution. This is the real dividing line. A bachelor thesis can largely apply existing methods to a question; a master's is expected to add something the field did not have, whether a new dataset, a fresh angle, a tested hypothesis or a synthesis nobody had written. Your introduction and discussion both have to name it.
• A defended method. It is no longer enough to say what you did. You have to argue why this method answers your question better than the alternatives.

How the months break down

A thesis runs in months, and the months have a shape. The split below is a working plan rather than a rule, and it exists to make one failure visible early: the single most common planning mistake is leaving nothing at the end, because formatting, the reference check and the defense all live in the final weeks and none of them compresses.

Two habits keep that plan from drifting. Book the supervisor meetings in advance rather than when you have something to show, because the meeting is what forces the something. And set targets a day can hold, one subsection rather than "work on the thesis", so the word count climbs without a heroic week in it.

Writing order, chapter by chapter

Do not write in document order. The introduction and abstract are easiest to write last, once you know what the thesis actually says, so most people start with the methodology, where you already did the work and are only describing it, or with the literature review, where the sources are in front of you. Momentum comes from starting where the writing is mechanical. Then write a rough version fast and fix it later, because a first draft that exists can be edited and a perfect paragraph in your head cannot.

One opinion, and it goes against the usual advice to finish the literature review before anything else. Draft the discussion, at least roughly, before the review is final. The review's job is not to survey a field, it is to set up the gap your results walk into, and you cannot aim it until you know what your results will say. Written first, it comes out as a competent tour of the reading with a gap bolted on at the end, and the discussion then reaches backwards for studies the review never introduced. Written second, it selects for the papers your discussion argues with, and the two chapters stop being strangers.

Whatever the order, keep the chapters talking to each other. The question you raised in the introduction is the one you answer in the conclusion, the gap from your literature review is the gap your method addresses, and your discussion reads your results against the studies you reviewed. A thesis that hangs together reads as one argument, not eight essays stapled together. Inside a chapter, the paragraph shape is the one any paper uses, claim first and then the evidence, which is step 6 of the research paper method.

A thesis also carries more citations than anything you have written, so pick the style your department requires on day one and use it everywhere. In Europe you will most often meet Harvard, APA, IEEE in engineering, or ISO 690. They are not interchangeable, and switching halfway through is a visible, avoidable mistake. Add each reference as you use the source rather than in a pass at the end, the habit that is step 7 of the method, and confirm each one exists before it enters the draft: search the title in Google Scholar, resolve the DOI at doi.org, and check the lead author publishes in the field. The full method is how to check if a citation is real, and the evidence on how often machine-produced reference lists fail that check is on can AI write my research paper.

Format, submit, defend

The last stretch is unglamorous and it is where marks quietly leak away. Format to your faculty's template: margins, font, line spacing, heading styles, page numbers, and the title page exactly as specified. Build the table of contents, check that every figure and table is numbered and captioned, and confirm that every in-text citation has a matching reference and every reference is cited. Then open the reference list and confirm each entry resolves to the paper you claim, which is the five-minute pass that is step 9 of the research paper method and the cheapest insurance in the document. This pass is also your last chance to catch borrowed phrasing you never put in your own words, so it is worth knowing how to avoid plagiarism before you submit.

Proofread properly, which at this length means more than one pass, and take the passes in the order that step 8 sets out: argument first, sentences second, typos last, ideally with a day between them. Reading aloud catches clumsy phrasing nothing else will. If your program offers a writing centre, this is what it is for.

Many programs end with a defense, sometimes called a colloquium, where you present the thesis and answer questions. It sounds frightening and it is usually fine. You know this document better than anyone in the room, because you wrote every word of it and made every decision in it. Prepare a short, clear presentation of your question, method and findings, anticipate the three questions that nearly always come (why this method, what are the limitations, what would you do differently), and practise out loud at least once. Calm preparation beats memorised lines, and a defense is mostly a conversation about work you did.

What stays yours

If the thesis feels overwhelming, that is mostly an illusion of scale. You are not writing a book this week. This week you are narrowing a question, or drafting one subsection, or verifying ten references. Do the step in front of you and trust that the steps add up, because they do.

CiteOwl helps with the parts that eat the months without touching the parts that have to be yours. It can structure your chapters, find and cite real sources for the literature review, and draft sections that arrive as plain diffs you accept or reject line by line, so the words stay yours. Every claim it writes links to a paper it actually read, with the supporting quote shown, so nothing fabricated slips into a reference list an examiner will spot-check. When you are done, it exports to the style your program requires.

What it does not do is the part you are being examined on. The question, the method, the argument and the defense are yours, and the record that proves it is a document whose citations resolve and whose edits you reviewed one by one. How far a tool can honestly carry a thesis is answered in full in can AI write my thesis, which walks through the declaration you sign, the supervisor who watched, and the room at the end.