A conference abstract is the short summary you submit to win a talk, poster or panel slot at an academic or professional meeting. It does two jobs. First it persuades a review committee that the work is sound, new and right for the program. Then, printed in the program or the app, it tells attendees whether your session is worth their hour. Each call sets its own length, from a 250-word paragraph to a 1,500-word short paper, and some meetings count characters instead of words.
Below you will find the five moves every abstract makes, the limits recent calls actually set, five complete examples from different fields with a note on why each works, a vague draft taken apart, and a six-step method. The studies, data and people in the examples are invented. The call you answer sets the real format, so read it before you borrow a structure from here.
What a conference abstract has to do
Almost every good abstract makes the same five moves, in the same order, whether or not the call asks for headings. Scientific abstracts follow the IMRaD pattern of the paper behind them. Humanities abstracts swap the method and results for a corpus and an argument, but the logic holds: here is a problem, here is what I did about it, here is what I can now tell you.
| Move | What it answers | In 250 words |
|---|---|---|
| MoveContext and gap | What it answersWhy does this matter, and what don't we know? | In 250 words40–50 words |
| MoveAim | What it answersWhat did you ask or set out to do? | In 250 words20–30 words |
| MoveMethod or material | What it answersHow, with what, and how many? | In 250 words40–60 words |
| MoveResult or argument | What it answersWhat did you find, or what do you claim? | In 250 words80–100 words |
| MoveImplication | What it answersSo what, and for whom? | In 250 words25–40 words |
A conference abstract differs from the abstract at the top of a journal article in three ways. It is often written months before the talk, sometimes before the analysis is finished. It is judged against a theme and a set of tracks, so fit matters as much as quality. And it is read in a hurry, by a committee working through a stack of submissions and then by attendees scrolling a program. Some meetings also publish their abstracts: the American Association for Cancer Research (AACR) issues its meeting abstracts as proceedings of the journal Cancer Research, so the text has a life after the conference.
Word limits vary by conference
There is no standard length. A public health meeting may give you 250 words, a cancer meeting a character count, a digital humanities conference up to 1,500 words for a long paper. The table shows what three recent calls asked for.
| Meeting | Field | Limit | What counts |
|---|---|---|---|
| MeetingAPHA Annual Meeting 2026 | FieldPublic health | Limit250 words | What countsAbstract text; some sections set their own limit |
| MeetingAACR Annual Meeting 2027 | FieldCancer research | Limit2,600 characters | What countsTitle, body and tables; spaces and authors excluded; each table counts as 800 |
| MeetingDH2026 (ADHO), poster | FieldDigital humanities | Limit500–750 words | What countsReferences excluded |
| MeetingDH2026 (ADHO), short paper | FieldDigital humanities | Limit750–1,000 words | What countsReferences excluded |
| MeetingDH2026 (ADHO), long paper | FieldDigital humanities | Limit1,250–1,500 words | What countsReferences excluded |
Character limits catch people out. A limit of 2,600 characters without spaces works out to roughly 400 to 500 words of prose, and a single table at AACR uses up almost a third of it.
The length also tells you what kind of abstract the committee wants. The University of North Carolina Writing Center's rule of thumb is that a limit of 100 to 120 words suggests a descriptive abstract (what the work covers), while 250 or more suggests an informative one (what it found). At the long end, the DH2026 lengths are closer to short papers, with room for an argument, its evidence and references.
Examples from the sciences
Empirical fields want results in the abstract, with numbers. These two examples show the two shapes you will meet most often: a single paragraph, and a structured abstract with headings.
Example 1: a results-first paragraph
Tree canopy cuts afternoon surface temperatures on elementary schoolyards by up to 19°C
Schoolyards in many US cities are paved, treeless and in use during the hottest hours of the day, yet few districts measure how hot they get. We asked how much tree canopy and surface material explain the differences in afternoon surface temperature across schoolyards in one mid-sized city. Between June and August 2025, we took thermal images of 40 public elementary schoolyards at 2 p.m. on 12 clear days and classified every surface as asphalt, rubber, artificial turf, grass or bare soil, in shade or in sun. Unshaded artificial turf was the hottest surface (mean 68°C), followed by unshaded asphalt (61°C) and rubber play surfaces (59°C). Shaded asphalt averaged 42°C. Canopy cover alone explained 58% of the variation in yard-wide mean temperature; surface material added 11%. Yards in the lowest-income quartile of attendance zones had a third of the canopy of yards in the highest. Because resurfacing costs more than planting and trees take years to grow, we recommend shade structures now and planting for the long term at the hottest yards. We will share the dataset and a one-page measurement protocol that any district can repeat.
Why it works. The gap takes two sentences and is specific: districts don't measure this. The method fits in one sentence with its numbers (40 yards, 12 days, one time of day). Half the abstract is results, each with a figure, and the equity finding gives the committee a reason to put it in front of policymakers. The title states the finding, so someone skimming the program learns something even if they skip the session.
Example 2: a structured clinical abstract
Two-way text reminders and missed visits in an outpatient diabetes clinic
Background: Missed appointments delay care and waste clinic time. In 2024, 23% of scheduled visits at our adult diabetes clinic were missed, and one-way text reminders had not moved that rate in two years.
Methods: Quality improvement study with a comparison clinic. From January to June 2025, patients received a text three days before each visit and could reply C to confirm or R to reschedule; a scheduler called back every R reply within one working day. A sister clinic with a similar patient mix kept one-way reminders. The primary outcome was the no-show rate, compared with July to December 2024 in a difference-in-differences analysis.
Results: Across 4,812 visits in the two clinics, the no-show rate fell from 23.1% to 15.4% in the intervention clinic and from 22.6% to 21.0% in the comparison clinic (difference in differences, −6.1 percentage points; 95% CI, −8.0 to −4.2). Staff refilled 64% of the slots freed by reschedule replies. The effect was smallest among patients over 75 (18.2% to 16.9%), who were also the least likely to reply.
Conclusions: Two-way reminders cut no-shows by about a third for roughly two hours of scheduler time a week. Older patients may need a phone option, which we are now testing.
Why it works. The four headings mirror what clinical calls usually request, so a reviewer finds the outcome in seconds. The comparison clinic answers the obvious objection (no-shows were falling anyway) before anyone raises it, and the confidence interval shows the effect is not noise. Giving the cost in staff time answers the first question a clinic manager in the audience will ask. Reporting the weak result for older patients reads as honesty, not as a flaw, and it sets up the next study.
When the results aren't in yet
Calls close months before the meeting (AACR's deadline for its April 2027 meeting is November 10, 2026), so many abstracts describe work in progress. That is acceptable in most social science and education meetings, as long as you say exactly what is done, what you have so far, and what the audience will get. Some biomedical meetings expect data in the abstract itself, so check the call first.
"I fixed the commas": how first-year students decide which feedback to act on
Instructors spend hours writing comments on draft essays, but we know little about how students choose which comments to act on. This study follows 24 first-year students in four sections of a required writing course at a public university through one drafting cycle. Students recorded think-aloud sessions while revising, and we matched each change in the final draft to the comment that prompted it, if any. Data collection ended in May 2026, and coding is complete for the first 12 students. So far, students acted on 71% of comments about grammar and citation format but only 38% of comments about argument or organization, and they most often skipped comments phrased as questions ("What is your evidence here?"). By the conference all 24 cases will be coded, and the presentation will report the full pattern with excerpts from the transcripts. We will close with three changes to comment wording that instructors can test in their own courses next term.
Why it works. It is honest about status (12 of 24 cases coded) and still gives the committee real numbers to judge. The promise at the end is specific and deliverable. Compare that with "findings will be discussed", which asks reviewers to accept a talk on trust. If you cannot yet state a research question this precisely, the abstract is not ready, whatever the deadline.
Humanities and practice abstracts
In the humanities the abstract usually proposes a paper that does not exist yet. There is no result to report, so the claim does that work: a reviewer should be able to disagree with it. Practice abstracts, common at teaching and professional meetings, describe something you did and what happened, and they promise materials people can take home.
Example 4: a paper for a literature panel
The typist as editor: Mina Harker's archive in Dracula
Dracula is usually read as a contest between a vampire and a band of men, but the book we hold is the product of someone else's labor. Mina Harker transcribes Seward's phonograph diary, types and copies the group's journals and letters, and arranges them in date order, turning private records into the dossier the men use to hunt the Count. This paper argues that Stoker makes compilation, not authorship, the decisive skill of the modern professional, and that he gives that skill to a woman while the plot keeps shutting her out of the hunt. Reading the novel beside typing manuals and office guides of the 1890s, I show how its documentary form registers a new kind of clerical work: anonymous, gendered and indispensable. For a panel on media and authority, the paper proposes that the novel's most important technology is not the phonograph or the telegraph but the filing system.
Why it works. It names the usual reading in one clause and then turns against it, which is what makes a humanities claim arguable. The evidence is concrete (what Mina actually does in the novel) and the method fits in one sentence. The last sentence ties the paper to the panel's theme in the panel's own words, which is the first thing a panel organizer checks.
Example 5: a teaching poster
Replacing a final exam with a public data story in introductory statistics
In spring 2026 we replaced the final exam in a 120-student introductory statistics course with a public project: teams of four found a local dataset, analyzed it, and published a two-page data story on the course website. Of 31 teams, 27 used city data, most often parking citations, library checkouts or restaurant inspections. Scores on the department's common reasoning questions were similar to the previous year's exam cohort (74% vs. 72%), while the share of students who said they could explain a confidence interval to a friend rose from 41% to 66% on the end-of-term survey. Teams struggled most with choosing a question their data could answer. The poster includes the assignment sheet, the rubric and a list of open city datasets that worked well.
Why it works. Practice abstracts are judged on whether others can reuse the idea. This one says what changed, gives one outcome measure and one perception measure, admits the hard part, and lists the materials on the poster. A teacher scanning the program knows exactly why to stop by.
How to write yours
Most weak abstracts fail the same way: they describe a topic instead of a finding. Here is the study from Example 1 as it might look in a first draft.
Heat is becoming an increasingly important issue for schools across the country as climate change continues to affect communities. This study explores the relationship between schoolyard design and temperature. Using a mixed-methods approach, we examine various factors that may contribute to heat on school grounds. Implications for policy and practice will be discussed, and recommendations will be presented at the conference.
- The first sentence could open a hundred papers. Replace it with the specific gap: districts don't measure schoolyard heat.
- "Mixed-methods approach" and "various factors" hide what was measured. Say thermal images, 40 yards, 2 p.m.
- "Will be discussed" withholds the result, and a committee cannot score what it cannot see. Give the numbers.
- There is no title-worthy finding. Once there is one, the title writes itself.
Example 1 is the same study after those four fixes. To get there from a blank page, work in this order.
- Read the call twice
Note the limit and what counts toward it, the formats on offer (talk, poster, panel), the theme and tracks, any required headings, whether review is anonymous, and how many abstracts you may submit. SfN's Neuroscience 2026 meeting, for example, allowed each member one scientific abstract, plus one in its history, education and society theme.
- Write the finding in one sentence
Before drafting, finish this sentence: "We found that…, which means… for…". In the humanities: "I argue that…, because…". If you cannot, you are writing a work-in-progress abstract, and Example 3 is your model.
- Draft long, one paragraph per move
Write the five moves without counting words. It is easier to cut a 400-word draft that says everything than to pad a thin one.
- Cut to the budget
Cut background first, then method details a reviewer does not need to judge the result. Keep the numbers. Spell out each abbreviation on first use; SfN's rules require it, and every reader benefits.
- Write the title last
Put the finding or the claim in it. A colon and subtitle earn their place only if they add information, such as the method or the population.
- Test it on an outsider
Ask a colleague from a neighboring field to tell you, after one read, what you found and why it matters. If they cannot, the problem is usually the result sentence, not the background.
Drafting and cutting in Strokes
An abstract goes through more versions than its length suggests: the long draft, the cut to the limit, the version a co-author rewrote. In Strokes you can keep the call's requirements, the full paper and every abstract draft as documents in one project on your machine. Select the draft and press ⌘K to ask for a version that keeps all the numbers and drops the background, or ask the AI Assistant whether the conclusion claims more than the results support. Version history saves the 400-word draft automatically, so the sentence you cut on Tuesday is a click away when a co-author asks for it back on Thursday.
Check every figure in an AI-shortened version against your data. Cutting is exactly where a qualifier disappears or a number changes, the kind of slip known as a hallucination, and you are responsible for every claim. Check the call's policy on AI as well: AACR's 2027 call asks authors to disclose AI-generated text in a sentence at the end of the abstract body.
If the abstract is the start of a longer paper, a research paper outline turns the five moves into sections.
Frequently asked questions
How long should a conference abstract be?
As long as the call allows, and no longer. Recent calls range from 250 words at the APHA Annual Meeting to 2,600 characters at AACR and up to 1,500 words for a long paper at DH2026. If the call gives no limit, ask the organizers rather than guess.
Can I submit an abstract before I have results?
In many fields, yes, if you are honest about it. Say what is done, give any preliminary numbers, and state what the session will report. Some biomedical meetings expect data in the abstract itself, so check the call.
How is a conference abstract different from a journal abstract?
A journal abstract summarizes a finished, peer-reviewed paper. A conference abstract is a proposal: it is often written before the work is complete, judged on fit with the program's theme, and read by attendees deciding which sessions to attend.
Should a conference abstract include references?
Short abstracts rarely have room for them, and many calls do not want them. Longer formats sometimes allow references and exclude them from the count, as DH2026 did. Follow the call.
Can I use AI to help write my abstract?
Check the call's policy first. Some meetings require disclosure: AACR's 2027 call asks for a sentence at the end of the abstract body when AI produced text. Whatever the policy, verify every number and claim yourself.
