How to Write an Abstract

More people will read your abstract than will ever read your paper – and most will decide from it alone.

An abstract is a self-contained summary of an entire piece of research, usually between 150 and 300 words. It sits at the front of a dissertation, a journal article or a conference submission, and it does most of the work of getting the paper read at all. Databases index it, reviewers screen on it, and researchers decide from it whether to open the full text.

Write it last

You cannot summarise findings you have not finished producing. Draft the abstract only when the paper is complete, working from the finished text rather than from your intentions. This also gives you a diagnostic: if the abstract is hard to write, the paper’s argument is probably unclear, and the difficulty is telling you where.

Cover the five things a reader needs

Whatever the discipline, an effective abstract answers the same questions in roughly the same order.

• Context and problem – what gap or question the work addresses, in a sentence or two.
• Purpose – what the study set out to do.
• Method – design, sample and approach, compressed but specific.
• Results – what you actually found, with key figures.
• Conclusion – what it means and why it matters.

The proportions are not those of the paper. Methods might occupy a third of the dissertation and two sentences of the abstract.

Give the findings, not a promise of findings

The most common failure is an abstract that describes what the paper will discuss without revealing what it concluded. “This study examines the relationship between sleep quality and academic performance” tells a reader nothing they could not guess from the title. “Sleep quality predicted exam performance independently of study hours, with each additional hour of disrupted sleep associated with a 4% decline in scores” gives them the finding, the size and the reason to read on.

Include real numbers. Effect sizes, sample sizes and percentages carry more information per word than any amount of description.

Know which format is expected

Structured abstracts use explicit labelled sections – Background, Methods, Results, Conclusions – and are standard in medicine, nursing and much of the health literature. Unstructured or descriptive abstracts run as continuous prose and are more common in the humanities and social sciences. Journals and universities specify which they want, along with a word limit, and both are enforced strictly. Check before drafting rather than reformatting afterwards.

Make it stand alone

An abstract is read separately from the paper, so it cannot depend on anything the paper contains. That means no references to “Table 3” or “the model described above”, no citations except in the rare cases a journal permits them, and no abbreviations that are not defined on the spot. Undefined acronyms are the single most common reason an abstract fails to communicate to anyone outside the immediate subfield.

Compress without becoming vague

Word limits force hard choices, and the wrong response is to generalise. “Several factors were found to influence outcomes” saves words by destroying the information. Cut instead by removing throat-clearing – “It is important to note that”, “This paper aims to argue that” – and by tightening constructions rather than removing content. Most first drafts lose a quarter of their length without losing a single fact.

Use the keywords people will search

Databases index the abstract, so the terms a researcher in your field would actually type should appear naturally in it. This is not keyword stuffing – it is making sure the standard vocabulary of your subject is present rather than replaced by an idiosyncratic phrasing. If your topic has a conventional name, use it.

Test it before you submit

Give the abstract alone to someone who has not read the paper and ask them to tell you what you did and what you found. If they can, it works. If they ask questions the paper answers, the abstract is leaning on material they do not have. Finally, check it against the paper for accuracy – abstracts are often written in a hurry at the end, and a figure that no longer matches the results section is a genuinely damaging error.

Match the abstract to its purpose

A conference abstract, a dissertation abstract and a journal abstract are not interchangeable. Conference abstracts are often submitted before the work is complete and may describe anticipated findings, though they should still state a clear contribution. Dissertation abstracts are read by examiners who will read the whole thing, so they function as a map. Journal abstracts are read by strangers deciding whether to read at all, and must work hardest of the three. Write for the reader you actually have.

Keywords and the surrounding metadata

Most submissions ask for keywords alongside the abstract, and they are worth more thought than they usually get. Choose terms that a researcher would search rather than terms that describe your paper poetically, avoid repeating words already in the title where the system indexes both, and use the standard vocabulary of your field – including, where one exists, the controlled vocabulary such as MeSH terms in health research. Together the title, abstract and keywords determine whether your work is ever found.

Draft it by extraction

A reliable method is to pull one or two sentences from each major section of the finished paper — the gap from the introduction, the design from the methods, the principal finding from the results, the implication from the discussion — and then edit that collection into continuous prose.

This guarantees the abstract reflects the paper rather than your intentions, and it makes the word limit a matter of compression rather than invention.

Front-load the important information

Readers scanning search results read the first sentence or two and stop. An abstract that spends its opening on general background loses them before the contribution appears. Establish the specific gap in one sentence and move to what you did.

Match tense to section

Conventions are consistent across most fields: present tense for established knowledge and for what the paper argues, past tense for what you did and found. Mixing them within an abstract is a common and easily corrected slip.

Avoid the standard weaknesses

Several patterns recur and each costs a reader.

• Describing rather than reporting — “the implications are discussed” instead of saying what they are.
• Undefined acronyms, which are opaque outside the immediate subfield.
• Citations, which most journals disallow in abstracts.
• Vague quantities — “several participants”, “a number of trials”.
• Claims the results do not support.

Write for the search as well as the reader

Databases index the abstract, so the standard terminology of your field should appear naturally within it. This is not keyword stuffing — it is ensuring that a researcher using the conventional term for your topic can find your work. Where your field has a controlled vocabulary, prefer its terms to idiosyncratic phrasing.

Check it against the paper at the end

Abstracts are usually written last and in haste, which is exactly when a figure gets transposed or a finding described more strongly than the results warrant. Reread it beside the results section as a final step. An abstract that overstates the paper is the most damaging error in the genre, because it is the part most people read and the part they will remember.