Your abstract will be read perhaps a hundred times for every once the full paper is. It is what appears in search results, what a handling editor reads before deciding whether to send your work out, and what a busy researcher uses to decide whether to open the PDF. It deserves more time than most authors give it.
What should an abstract contain?
Five things, in most fields one sentence each: the problem, the gap, what you did, what you found, and what it means. If any of the five is missing, readers cannot place your work.
- Context — the problem, in a sentence a neighbouring field would understand.
- Gap — what is not yet known, or what previous work could not do.
- Approach — what you actually did, with the method named.
- Results — the finding, with numbers.
- Implication — why it changes anything.
How long should an abstract be?
Whatever the journal's author guidelines specify, usually between 150 and 300 words. Exceeding the limit is one of the few things that can get a submission returned before an editor reads it.
Where the journal requires a structured abstract with labelled headings — common in medicine — use exactly the headings given, in the order given.
Should you write the abstract first or last?
Last, always. An abstract written first describes the paper you intended to write, and the two are rarely the same by the end.
Draft it once the discussion is final, then check every claim in it against the results section. A claim the results do not quite support is one of the fastest routes to a sceptical reviewer.
Should an abstract include citations?
No, in almost every case — an abstract is read and indexed detached from the reference list, so a citation in it points at nothing. The rare exception is a paper responding directly to one specific prior study, and even then the journal may forbid it.
How do keywords affect who finds your paper?
They determine whether your paper appears in database searches at all, so choose the terms a stranger would use rather than the ones your research group uses internally. Avoid abbreviations unless they are universal in your field.
A practical test: search your own keywords in Scopus or Google Scholar. If your paper would not stand out among the results, they are too broad; if there are almost no results, too narrow.
What are the most common mistakes?
- Background that eats half the words. One sentence of context is usually enough.
- No numbers. "Significantly improved" tells a reader nothing; "improved by 23% (p < 0.01)" tells them everything.
- Undefined abbreviations. The abstract stands alone — define anything that is not universal.
- Overclaiming. Reviewers check the abstract against the results, and a mismatch damages trust in everything else.
- Sentences copied from the introduction. Reviewers notice, and it reads as haste.
How can you test whether it works?
Give it to a researcher one field away and ask them what you did and why it matters. If they can answer both in their own words, the abstract works; if they hesitate, the problem is your writing rather than their attention.
The abstract is one of the highest-leverage half-days in the whole publication process, which is why it is the first thing we look at when reviewing a manuscript. If you would like a fresh reader on yours, send it over.
