How to Write an Abstract (With Examples)

To write an abstract, summarize your finished paper in one tight paragraph that covers five things in order: the background and problem, your aim, your method, your key results, and your conclusion. Write it last, after the rest of the paper is settled, because an abstract is a summary and you cannot summarize what does not exist yet. Keep it self-contained, leave citations out, and stay inside the word limit your assignment or journal gives. Most abstracts run 150 to 300 words, but that number varies, so check your guidelines first. Get those five parts in and you have an abstract, whatever the field.

The abstract is the shortest section of your paper and the one most people read. It is the first thing a marker, a reviewer, or a database search shows, and for many readers it is the only part they ever see. That makes it strangely high-stakes for 200 words. The good news is that a strong abstract is almost mechanical once you know what goes in it: a fixed set of parts, in a fixed order, written about a paper you have already finished. This guide covers what an abstract is and why it comes last, the five parts to include, structured versus unstructured forms, length, keywords, a worked example, and the mistakes that quietly cost marks.

What an abstract is, and why you write it last

An abstract is a short, standalone summary of your whole paper, usually one paragraph, that lets a reader understand what you did and what you found without reading the rest. It sits at the top of the paper, right after the title, but it is not an introduction. An introduction sets up the problem and leaves the answer for later; an abstract gives away the whole story, including the results and the conclusion. Nothing is held back for suspense, because its job is to let a reader decide, in thirty seconds, whether the full paper is worth their time.

That self-contained quality is the rule that governs everything else. The abstract has to make sense on its own, pulled out of the paper and dropped into a search results page, which is exactly what databases do with it. So it carries no citations, no figure references, no "as discussed below," and no abbreviation a first-time reader would not recognize. If a sentence only makes sense to someone who has already read the paper, it does not belong in the abstract.

And you write it last. This is the single most common mistake: drafting the abstract early, when it feels like an outline of intentions. But an abstract summarizes a finished paper, and your aim sharpens, your method shifts, and your results are not final until the analysis is done. An abstract written first describes a paper that no longer exists by the time you submit. Write the body, the introduction, and the conclusion, and only then turn around and summarize what you actually wrote. We put the abstract at the end of the build for the same reason in how to write a research paper.

The five parts of an abstract

Whatever your field, a good abstract answers five questions in order. Think of it as one sentence or two per part, which is how 200 words holds an entire paper.

1. Background and problem

Open with the context and the gap. One or two sentences on why the topic matters and what is missing, unresolved, or contested in what is already known. This is not a literature review; it is the single problem your paper addresses, framed so a reader outside your exact subfield understands why anyone should care.

2. Aim

State what your paper set out to do. This is your research question or objective in one clean sentence: what you investigated, tested, or argued. The aim follows directly from the gap you just named, so the reader sees the problem and your response to it back to back.

3. Method

Say how you did it. For an empirical paper, that is your design, participants or data, and approach, briefly: enough for a reader to judge how the results were produced, not a full methods section. For a non-empirical paper, it is your approach, the framework, the texts, or the kind of analysis you ran. Keep it to the essentials that make your results credible.

4. Key results

Report what you found. This is the part students underwrite, hiding the actual findings behind "results are discussed." Give the real outcome: the main result, the direction of the effect, the headline number if there is one. The results are why anyone reads the abstract, so they get the most weight. Report what the paper found, not what you hoped it would.

5. Conclusion

End with what it means. One sentence on the takeaway: the answer to your aim, why it matters, or what it implies for the field or for practice. This is where you state the significance without overclaiming. A conclusion that says more than your results support is the fastest way to lose a reader's trust in the first paragraph.

Structured versus unstructured abstracts

The five parts stay the same; what changes is whether they wear labels. An unstructured abstract is one flowing paragraph with no headings, and it is the default for most humanities, social science, and undergraduate work. The five parts are still there, just stitched into prose. This guide's worked example below is unstructured.

A structured abstract breaks the same content into labeled sections, often Background, Methods, Results, and Conclusions, each as its own short block. Medical, health, and many science journals require this format, and some assignment briefs ask for it too. If yours does, follow the exact headings it specifies, because the labels are not optional decoration; they are the format. The content you write is identical either way. The only question is whether your guidelines want the seams showing.

Descriptive versus informative abstracts

There is a second distinction worth knowing, because the wrong one fails the assignment. A descriptive abstract says what the paper is about without giving away the findings: it names the topic, the scope, and the purpose, but stops short of the results and conclusion. It is short, often under 100 words, and you will mostly meet it in the humanities or as a conference proposal.

An informative abstract gives the whole picture, results and conclusion included, and it is what almost every research paper and journal expects. When an assignment says "write an abstract" with no other detail, it nearly always means the informative kind. If you write a descriptive abstract by accident, leaving out your findings, it reads as if you never reached any, which is the opposite of what you want. The five-part structure above is informative by design.

How long should an abstract be?

It varies, and the only number that matters is the one in your guidelines. Check the assignment brief or the journal's author instructions before you write, because the word limit is a hard cap and going over it can cost marks or get a paper desk-rejected. The typical ranges are still predictable. Most abstracts run 150 to 300 words. Student papers are often capped around 250, many journals sit at 200 to 250, and a conference abstract can be shorter. A structured abstract with labeled sections sometimes runs a little longer to fit the headings.

When no limit is given, aim for one tight paragraph that covers all five parts and nothing else. The discipline of a word count is useful, not annoying: it forces every sentence to earn its place and pushes the padding out. If you cannot fit all five parts in the limit, you are explaining too much, not summarizing too little. Cut the setup and keep the substance.

A worked example

Here is an informative, unstructured abstract for an invented student study, built from the five parts. The topic and findings are illustrative, not drawn from any real paper, so use it as a shape to copy, not a source to cite.

Read it back against the five parts. The first sentence is background and the gap (familiarity is understudied). The second is the aim. The third is the method, design and sample in one line. The fourth is the key results, with the actual direction of the effect rather than "results are reported." The last is the conclusion, stating what it means without claiming more than the study showed. No citations, no abbreviations, no reference to figures, and it stands completely on its own. That is the whole job.

Keywords

Many journals and some assignments ask for a short list of keywords, usually three to six, placed just under the abstract. They are not decoration; they are how your paper gets found, the terms a database indexes and a searcher types. Choose the words a reader looking for your work would actually use: the core concepts, the method, the population, the field. Skip words already in your title where you can, since those are indexed anyway, and prefer the standard term over a clever one. For the example above, "background music, reading comprehension, attention, undergraduates, familiarity" would do the job.

Common mistakes

Most weak abstracts share a short list of fixable problems:

• Including citations or reference markers. An abstract is standalone. Leave the citations, footnotes, and "[3]" markers in the body where they belong.
• Being vague about results. "Findings are discussed" tells the reader nothing. State the actual result, including the number or direction if there is one.
• Writing it first. An abstract drafted before the paper describes a paper you did not write. Save it for last.
• Treating it as an introduction. An intro withholds the answer; an abstract gives it away. Include your results and conclusion.
• Going over the word limit. The cap is a rule, not a suggestion. Cut the setup before you cut the findings.
• Unexplained abbreviations or jargon. The abstract reaches readers outside your subfield. Spell things out or leave them for the body.
• Claiming more than the paper shows. An overreaching conclusion undermines the work in its most-read sentence. Match the claim to the evidence.

Where CiteOwl fits

The abstract is the one part of the paper that is genuinely easier to automate, because it is a faithful summary of work you already did rather than an argument you have to think up. CiteOwl drafts the abstract from your finished paper, last, the way you should: it reads what you actually wrote and produces a summary that hits the five parts, and that draft arrives as a reviewable change you accept, edit, or reject like any other. It does not invent a finding that is missing from your results or smuggle in a citation, because the abstract is built from the paper in front of it, not from memory.

That is the same verify-first principle that governs the rest of the tool. CiteOwl searches actual literature and reads what it finds, so the sources behind your paper are real, and every change it makes, the abstract included, is a diff you review with version history to step back. You keep the argument and the final words; the tool handles the slow, mechanical parts and shows its work. If your next job is the survey of the field that feeds the paper, here is how to write a literature review with AI.