---
name: conference-abstract-and-presentation
description: >
  Writes conference abstracts that get selected and designs the talk, poster or
  lightning session that follows. Use for "write a conference abstract", "submit
  to a conference", "structured abstract", "my abstract was rejected", "prepare a
  conference talk", "design a research poster", "how do I present unfinished
  research", "my talk overruns", "prepare for the question session", "conference
  networking".
category: 15 Academic University Research
ref: "15.27"
tier: 3
inherits: [K2, K3, K4, K5]
---

# Conference Abstract and Presentation

## 1. One-line description

A method for writing a conference abstract that survives a competitive selection read of under two minutes, and for designing the talk, poster or lightning session that follows as a distinct format rather than as a compressed paper.

## 2. What this skill is used for

**The research problem it solves.** Conference submission and conference presentation are separate skills, and most researchers are trained in neither. An abstract is judged on two hundred to four hundred words by a reviewer working through a stack of them, who needs to know within a few sentences what was studied, what was found, and why this belongs in this track. Abstracts fail for reasons that have nothing to do with the quality of the work: no result stated, a promise that findings "will be discussed", the wrong track, an audience mismatch, or a claim that overreaches what the work has actually established. The presentation then fails differently. A talk is not a paper read aloud, because a listener cannot re-read a sentence, cannot skip back, and loses the thread permanently the moment the density exceeds what speech carries. A poster is not a paper printed large, because it is read standing up, from a distance, in noise, by someone deciding in three seconds whether to stop. This skill treats each format as its own design problem and supplies the method for each.

**Where it sits in the research lifecycle.** Alongside everything. Conference submission often happens well before a study is finished, which is normal in most fields and creates the specific honesty problem this skill addresses in Step 3. Presentation happens during the work, after it, and frequently before the journal article exists.

**Typical use cases.**
- Writing an abstract for a competitive conference call, structured or unstructured.
- Choosing between tracks, or working out why a submission was rejected.
- Deciding how to submit work that will not be finished by the conference date.
- Designing a talk from a paper, without compressing the paper.
- Designing a poster that works at conference-hall distance.
- Preparing for the question session, including the questions you hope nobody asks.
- Fitting a talk into its slot, which almost no academic talk does by default.
- Making a conference produce something afterwards, without pretending to be a different personality.

**Who uses it.** Doctoral researchers presenting for the first time; early-career academics for whom conference presence is a career necessity; established researchers submitting to an unfamiliar community; applied researchers presenting to practitioner audiences; anyone who has been told their talk overran.

## 3. When to use it

- A conference call is open and you are deciding what to submit and where.
- You have a draft abstract and no idea whether it is competitive.
- An abstract was rejected without informative feedback.
- Your submission was accepted and you now have to build the talk or poster.
- The work will not be finished by the conference and you need to submit honestly.
- You have been given a slot length and your material does not fit it.
- You are preparing for questions, or have been caught out by a question before.
- You are attending a conference and want it to produce something beyond attendance.

## 4. When NOT to use it

- **The work has no findings and no defensible expectation of having any.** Submitting an abstract for a study that has not begun, with results promised, is overclaiming, and in fields with any reporting standard it is a recognised problem. Where a study is genuinely at the design stage, submit it as what it is: a protocol, a method paper, a work-in-progress session, or a poster describing the design, all of which most conferences accept. See Step 3.
- **The purpose is to convert the paper into a journal article.** That is **15.24 Journal Article Development**. A conference abstract and a journal abstract are different instruments with different readers, and using one for the other fails in both directions.
- **The choice is between a conference and a journal as the primary venue for the work.** That is **15.25 Journal and Venue Selection**, which handles the field-dependent question of whether an archival conference is a terminal venue or a step. This skill assumes the conference decision has been made.
- **The presentation is a thesis defence or viva.** Different format, different audience, different stakes, and the preparation is adversarial rather than communicative. See **15.12 Thesis Defence and Viva Preparation** and **15.19 Examiner-Proofing and Defence Strategy**.
- **The audience is a client, a funder or a decision-making body.** A conference talk is built to inform peers; a stakeholder presentation is built to support a decision, and the structure, evidence standard and register all differ. See **12.03 Research Report Compilation** and the reporting track for that work.
- **Presenting would breach confidentiality, embargo or consent.** Commercially confidential data, an embargoed finding, participant material whose consent did not extend to public presentation, or images and quotations that could identify a person in a small community. Photography and recording at conferences are normal and mean anything shown is effectively public. Resolve first: **13.05 Research Ethics and Consent Design**. `RESEARCHER DECISION REQUIRED` (K5 §2.4).
- **What is wanted is a talk script or an abstract to submit as the author's own unaided writing.** See §12.1.
- **The abstract is for a venue whose selection is not competitive and whose purpose is attendance.** The effort here is calibrated to competitive selection. Where acceptance is effectively automatic, spend the effort on the presentation instead, and note that this is also a signal worth examining about the venue (**15.25 Journal and Venue Selection**, Step 7).

## 5. Required inputs

**Required.**
- **The call for submissions**, including the track list, the word limit, the abstract format (structured or free), the submission deadline, and the selection criteria where published. Abstracts are frequently rejected for exceeding a limit or missing a required field, which is the cheapest possible failure.
- **The work itself**, at whatever stage it has actually reached, stated honestly. Design only, data collected, analysis under way, or complete. This determines what can be claimed (Step 3).
- **The format accepted or sought**: full talk, short talk, lightning or flash session, poster, panel, workshop, symposium contribution. Each is a different design problem and a talk cannot be shortened into a lightning session.

**Optional, and what each one adds.**
- **The conference's previous programme.** The single most useful optional input. It shows what this community actually accepts, the level of completeness expected, the typical scope of a contribution, and which track hosts work like yours.
- **The slot length and room format.** Determines content volume directly, and whether questions are inside or outside the slot. A twelve-minute talk with three minutes of questions is a different object from a twenty-minute talk.
- **The audience composition.** Specialists in the subfield, the broad discipline, practitioners, students, or a mix. Determines how much is assumed and how the significance is framed.
- **The paper, if one exists.** Useful as source material, dangerous as a template. Step 5 exists because the paper's structure is the wrong structure for a talk.
- **Any co-presenters or co-authors**, and who is presenting. Determines rehearsal needs and who takes which questions.
- **Accessibility requirements and room constraints.** Screen size and aspect ratio, whether the room is bright, whether a microphone is used, whether captioning or interpretation is provided, and whether materials are shared in advance.

## 6. Questions to ask before starting

1. **Which track, and who reads that track's submissions?** Track fit is the most common avoidable rejection, and the reviewers differ by track. Default if unanswered: derive the track from the previous programme rather than from the track titles, which are broader than their contents.
2. **What state will the work actually be in on the conference date, not on the submission date?** Determines what the abstract may promise. Default: assume the work will be no further along than a realistic projection, and write the abstract so it remains true if nothing further is completed (Step 3).
3. **What format, and how long is the slot?** Determines content volume before any content is chosen. Default: ask; do not design a talk before knowing its length, because the length is the primary constraint.
4. **Who is in the room?** Determines assumed knowledge. Default: assume a mixed audience with a specialist minority, and design so the specialists are not bored and the rest are not lost.
5. **Is this abstract published, indexed or citable?** Some conferences publish abstracts as a permanent record, which raises the accuracy standard and creates a prior-publication question for the eventual paper. Default: assume it is permanent and check.
6. **What do you want to have happened by the end of the conference?** Determines whether the effort goes into the talk, the poster, the follow-up, or the sessions attended. Default: assume the talk, and raise the question, because most people never ask it.
7. **What is the one question you hope nobody asks?** The fastest route to the weakest point in the work. Default: ask it directly; the answer is almost always immediate.

## 7. Step-by-step methodology

**Step 1. Understand the selection you are writing for.**
An abstract reviewer is typically a volunteer working through twenty to a hundred submissions, giving each a couple of minutes, scoring against criteria that usually include relevance to the conference and track, contribution or novelty, methodological soundness as far as the abstract shows it, and clarity. They cannot check anything, cannot ask you a question, and have only the text. Everything that follows is a consequence: the first two sentences carry disproportionate weight, a reviewer who has to work out what you did scores you lower than one who is told, jargon from a neighbouring subfield reads as unclear rather than expert, and an abstract with no stated result gives them nothing to score on contribution. Where the criteria are published, read them and check each is addressable from your text; where they are not, the four above are a safe default. *Correct result: the criteria written down, and a plan for which sentence in the abstract satisfies each.*

**Step 2. Write the structured abstract's fields, even where a structured abstract is not required.**
Structured abstracts ask for explicit fields, typically some version of background or objective, methods, results and conclusion, with variations by field (a design field may ask for the problem and approach; a humanities conference for the argument and its sources). Where they are required, fill each field with what it asks for and nothing else, since a reviewer scanning for the method should find it under the method heading. Where they are not required, write the fields anyway as a drafting device and then dissolve them into prose, because free-form abstracts fail predominantly by omission and the fields make omission visible. The four contents, whatever they are called: **why this matters and what question it addresses** (one or two sentences, and where the reviewer decides whether to care); **what you actually did**, specifically enough to judge, including design, participants or materials, and analysis; **what you found**, with the actual finding; and **what it means**, in one sentence that is not a promise of discussion. Word counts are tight, and the part that must not be squeezed is the result. *Correct result: a draft in which each of the four contents is present and identifiable, within the word limit.*

**Step 3. Handle unfinished work honestly, which is the normal case and not a problem.**
Most conference deadlines fall months before the conference, and submitting work in progress is standard practice in most fields. The failure is not incompleteness; it is misrepresentation. Three honest forms exist and one dishonest one. **Report what is complete**, confining the abstract to the finished part, which is strongest where a subset of the work stands alone. **Describe the design and the planned analysis, and say the results will be presented**, which is legitimate where the track accepts work in progress and the abstract is explicit that data collection or analysis is under way. **Submit to the format built for it**: work-in-progress tracks, poster sessions, doctoral consortia and lightning sessions exist precisely for this, and are often less competitive and better suited. The dishonest form is stating or implying results that do not yet exist, including through passive constructions that make an expectation look like an outcome ("the intervention was found to improve..." written before the analysis ran). Never write a result you do not have (K4 §2.1). If the study then fails to produce the expected finding, the abstract has become a public commitment you must break or fudge in front of an audience, and a published abstract makes that permanent. Where the abstract is honest about being in progress, say precisely what will be complete by the conference date, and be conservative, because that estimate goes wrong in one direction far more often than the other. *Correct result: an abstract that will still be accurate if no further work is completed, and that says clearly what stage the work has reached.*

**Step 4. Match the submission to the track and the community, and check by reading the previous programme.**
Track descriptions are broad and overlapping; the previous programme shows what each track actually contains. Read one and place your work: if you cannot find three talks it would sit beside, either the track is wrong or the conference is. This also calibrates the two things submitters most often get wrong: the expected size of a contribution (some communities accept a single well-executed study, others expect a programme of work) and the expected completeness. Note too the register and the mix of theory, method and application in typical contributions, because a submission pitched at the wrong mix reads as an outsider's even when the work is strong. Where the audience extends beyond the immediate subfield, frame the significance to reach them without diluting the specificity that makes it credible. *Correct result: a track choice justified by named prior contributions it would sit beside.*

**Step 5. Design the talk as a talk, not as a compressed paper.**
The governing fact is that **a listener cannot re-read**, and every consequence follows from it. A listener who loses the thread does not recover it, so the structure must be announced, simple and repeatedly signposted. Density that is comfortable on a page is impossible in speech, so a talk carries roughly a quarter of the content of its paper. And a listener retains a small number of things, so the talk must decide in advance what those are. The method: choose **one message** the audience will still have in a week and write it as a sentence; then choose the smallest number of things that must be true for it to land, usually three to five; then build the talk as what the problem is and why the room should care, what you did in enough detail to be credible and no more, what you found, what it means, and what is next. Method gets less time than researchers expect and results get more, because the audience is trusting your competence in the room and interrogating your findings. Cut everything not load-bearing for the message and hold it in reserve for questions, where it is far more valuable. Signpost transitions explicitly, because the audible structure is the only structure a listener has. *Correct result: a one-sentence message, three to five supporting points, and a content list that is a fraction of the paper.*

**Step 6. Build the visual material with one idea per slide and enough contrast to survive the room.**
Slides support speech; they do not duplicate it. A text-dense slide makes the audience read instead of listen, and reading beats listening, so you lose them. The discipline that does most work is **one idea per slide**, which forces the content decisions to happen at design time rather than in front of the audience. Practically: few words, large type, one chart per slide with its point stated as the slide title rather than a neutral label, axes and units legible from the back row, no table a listener would have to read, and no figure imported at paper scale, which is the commonest visual failure and is unreadable in every room. Colour must survive projection in a bright room and must not be the only carrier of meaning, both for legibility and because a proportion of any audience will not distinguish the hues you chose. Give caveats verbally rather than as small print nobody reads. Roughly one slide per minute is a reasonable first estimate for straightforward slides, and it is an estimate to check against the clock, not a target to fill. *Correct result: slides that would be intelligible from the back row, each carrying one point, with the point stated rather than implied.*

**Step 7. Design a poster for the reader who is standing three metres away and deciding whether to stop.**
A poster is read at three distances by three different readers and must work at all three. From three metres, the title and one central visual must convey the topic and the finding; this is the only chance to make a passer-by stop, so state the finding in the title where the field permits. From one metre, the structure must be scannable: few clearly separated sections, an obvious reading order, the key result visually dominant. From close up, for the one reader in twenty who reads everything, detail must be present but subordinate. The consequences are unambiguous: much less text than a paper (a poster with a paper's word count is a paper on a wall, and is not read), large type with body text legible at a metre, generous empty space, one dominant graphic, and the finding stated in words large enough to read from the aisle. Prepare separately the thirty-second spoken summary you will give repeatedly, because the session is a conversation and the poster is its prop. Bring something to leave: a link, a short handout, a way to be contacted. *Correct result: a poster whose finding is readable from three metres, whose structure is obvious from one, and a rehearsed thirty-second summary.*

**Step 8. Design a lightning or flash session as its own format.**
A lightning session (typically three to five minutes, sometimes with a fixed slide count or automatic advance) is not a short talk and cannot be produced by compressing one. It carries exactly one idea, and its job is not to convey the study but to make the right people want to talk to you afterwards. Build it as: the problem in one sentence, the thing you found in one sentence, one piece of evidence, and an invitation to continue elsewhere. There is no room for method detail, and attempting it produces the characteristic failure of the format, which is a talk delivered at high speed that nobody follows. Where the format enforces automatic slide advance, rehearse against the clock without exception, because there is no recovery from falling behind. *Correct result: one idea, four or five slides, rehearsed to the second, ending with a reason to find you.*

**Step 9. Rehearse against a clock, and cut, because almost every academic talk overruns.**
Overrunning is close to universal and is treated as a minor sin when it is a serious discourtesy: it takes time from the next speaker, truncates the questions, and where a chair intervenes it costs the speaker the conclusion, which is the part that mattered. The causes are structural rather than personal. Speakers estimate rather than time; they speak more slowly under pressure than in rehearsal, or faster and less intelligibly; they add unrehearsed explanation when a slide looks thin; and they build for the full slot, leaving no margin. The remedy is mechanical: rehearse aloud, standing, with the slides, against a clock, at least twice, timing each section. Build for eighty per cent of the slot, leaving margin for the slow start, the technical problem and the clarifying question. Know in advance which two slides you will drop if you fall behind, and mark them, so the cut is a calm decision beforehand rather than a panic in the room. Rehearse the opening and closing sentences until they are secure, because the opening sets the register, the closing is what the room takes away, and both are the parts most likely to be lost to nerves. Do not read a script, which is audibly different from speech and loses a room quickly, but do write out the two or three sentences that must be exactly right. *Correct result: a timed run that finishes with margin, a marked cut list, and a secure opening and close.*

**Step 10. Prepare the question session, which is the part that is actually assessed.**
For many audiences the questions reveal more than the talk, and preparation is straightforward and almost nobody does it. List the ten questions you expect, including the obvious ones on sample, generalisability, method choice and the alternative explanation, and prepare a thirty-second answer to each. Then list the questions you hope nobody asks, which identify the real weaknesses, and prepare honest answers, because the honest answer to a known weakness ("that is a limitation, here is why it does not undermine the specific claim, and here is what would settle it") is far stronger than an improvised defence. Prepare the answer to "what is the sample size" or its field equivalent, which will be asked. Have the material cut from the talk ready as backup slides, which is the most efficient possible use of it. Then the conduct: listen to the whole question rather than answering the first half, restate it where it was long or unclear, answer briefly, and say "I do not know" where you do not, which is a normal professional answer heard as confidence rather than weakness. Where a questioner is hostile or making a speech, answer the substantive part civilly, decline to escalate, and offer to continue afterwards. Where a question exposes something you had not considered, say so; it is the most credible thing you can do. *Correct result: ten prepared answers, the feared questions among them, backup slides ready, and an agreed plan for who answers what where there are co-presenters.*

**Step 11. Plan what the conference should produce, in practical terms.**
Advice to "network more" is useless to the people who most need it, because it prescribes a personality rather than an action. Replace it with a plan. Before: identify five to ten people whose work you actually engage with (they are in your reference list, which is where to look), read something recent by each, and write one specific question you would genuinely ask them. Pick the sessions you will attend for reasons other than obligation. During: ask one question in a session, which is the lowest-cost way to become someone whose name others recognise and is easier from the audience than from the front. Talk to the person at the poster next to yours, since proximity does the introduction for you. A short, specific comment on someone's work is a complete interaction and need not become a conversation. After: send two or three short messages within a week referring to the specific thing discussed, since a message sent later to someone who no longer remembers you is what makes people conclude networking does not work. None of this requires being outgoing; it requires a list, some preparation and a few specific actions, which is a much more tractable problem. *Correct result: a named list with a prepared question each, a session plan, and a follow-up commitment with a deadline.*

## 8. Analytical framework

Two frameworks, one for the submission and one for the presentation.

**The abstract is scored on four things, and each needs a sentence:**

    Relevance (why here, this track, this audience)
      → Contribution (what is new or useful)
        → Soundness (what you did, specifically enough to be judged)
          → Clarity (can a tired reviewer follow it once, quickly)

The dependency runs left to right: a reviewer who cannot establish relevance stops reading, and clarity failures make the other three unassessable. The most common structural fault is an abstract that spends half its words on relevance and has none left for contribution.

**The presentation is designed from the constraint inward:**

    Format and slot length → One message → Supporting points (3 to 5)
      → Evidence for each → Visual per point → Time per section → Cut list

Design in this order, never the reverse. Starting from the material and asking what will fit produces a compressed paper, which is the failure the whole method exists to prevent. Starting from the constraint and asking what earns its place produces a talk. The cut list at the end is not a leftover: it is the reserve that answers the questions.

## 9. Output format

**1. Submission package.** Track, format, word limit, deadline, and the selection criteria with the sentence in the abstract that addresses each.

**2. The abstract**, structured to the required fields, or in prose with the four contents present and identifiable. Word count stated.

**3. Completeness statement.** What stage the work has actually reached, what will be complete by the conference date, and confirmation that the abstract remains accurate if nothing further is completed.

**4. Presentation design.**

| Element | Content |
|---|---|
| Format and slot | |
| One message (one sentence) | |
| Supporting points (3 to 5) | |
| Evidence per point | |
| Time budget per section (to 80% of slot) | |
| Cut list, in cut order | |
| Backup slides for questions | |

**5. Slides or poster**, built to the Step 6 or Step 7 discipline, with the readability check recorded.

**6. Question preparation.** Ten expected questions with prepared answers, the feared questions among them, and who answers what where there are co-presenters.

**7. Conference plan.** People, sessions, and the follow-up commitment with its deadline.

**When results do not exist yet, the format must say so rather than fill the slot (K4 §1).** The results field of a structured abstract for work in progress states the analysis planned and the stage reached; it does not contain an anticipated finding. In the talk, a slide reporting an incomplete analysis says what has been done and what has not. The correct output for a study that has produced nothing is a submission describing the design, not a submission describing hoped-for outcomes.

## 10. Quality checks

Run before submitting and before presenting. These sit on top of K4 §8.

1. Does the abstract state an actual result, or only that results will be presented?
2. Is every claim in the abstract true of the work as it stands today?
3. Would the abstract still be accurate if no further work were completed before the conference?
4. Is the track choice evidenced by named prior contributions the work would sit beside?
5. Is the abstract within the word limit and does it fill every required field?
6. Could a reviewer outside your immediate subfield follow the abstract on one reading?
7. Does the talk have one message, stated as a sentence, that a listener would still have in a week?
8. Is the talk designed from the slot length, or compressed from the paper?
9. Does any slide carry more than one idea, or text a listener would have to read?
10. Is any figure imported at paper scale, and is every axis legible from the back row?
11. Does colour carry meaning that is lost if the hues are not distinguished?
12. Has the talk been timed aloud, standing, with slides, and does it finish inside eighty per cent of the slot?
13. Is the cut list decided and marked in advance?
14. Is the poster's finding readable from three metres, and is its reading order obvious?
15. Are the ten expected questions prepared, including the ones you hope nobody asks?
16. Does anything being shown breach an embargo, a confidentiality condition or the consent given (K5 §2.4)?

## 11. Common failure modes

| Failure | How to recognise it | How to prevent it |
|---|---|---|
| **"Results will be discussed"** | The abstract promises rather than reports | Report what exists; use a work-in-progress format if that is what it is (Steps 2 and 3) |
| **Promised results that do not arrive** | The abstract states a finding the analysis had not produced | Never write a result you do not have (Step 3, K4 §2.1) |
| **Wrong track** | Rejected without substantive feedback | Place the work against the previous programme (Step 4) |
| **All context, no contribution** | Half the words spent on background | One or two sentences of relevance, then the contribution (Step 2) |
| **Subfield jargon** | Reviewers outside the niche cannot follow it | Write for the track's actual readership (Steps 1 and 4) |
| **The compressed paper** | The talk has the paper's section structure | Design from the slot inward (Step 5) |
| **Wall-of-text slides** | The audience is reading, not listening | One idea per slide, few words (Step 6) |
| **Paper figure on a slide** | Six-panel figure, unreadable axes | Rebuild each figure for the room (Step 6) |
| **The poster that is a paper** | Paper-length text on a wall; nobody stops | Three-metre, one-metre, close-up design (Step 7) |
| **The compressed lightning talk** | Five minutes of high-speed method | One idea and an invitation (Step 8) |
| **Overrunning** | The chair cuts you off before the conclusion | Rehearse against a clock, build to 80% (Step 9) |
| **Unprepared questions** | An improvised defence of a known weakness | Prepare ten answers, including the feared ones (Step 10) |
| **Answering half the question** | The questioner repeats themselves | Listen to the end, restate it, then answer (Step 10) |
| **Networking as personality advice** | "Be more outgoing", no action taken | A named list, a prepared question each, a follow-up deadline (Step 11) |
| **AI-inflated abstract** | Confident phrasing about a modest or unfinished study | Check every claim against the work's actual state (§12.2) |

## 12. AI guardrails

Skill-specific. The universal prohibitions in K4 apply in full and are not repeated.

1. **Academic integrity is the binding constraint on everything in this skill.** These skills assist a researcher's thinking, structure and rigour. They do not produce work to be submitted as the researcher's own unaided output. The user must comply with the policies of their institution and of the venue they are submitting to, including any requirement to declare AI assistance. Many publishers now require disclosure of AI use in manuscript preparation and prohibit it entirely in peer review. The skill never writes a passage for submission as though the author wrote it; it interrogates, structures, critiques and teaches. Conference abstracts are submissions and many conferences are now explicit about AI use in them; check the call and comply.

2. **Never write a result, an effect, a participant count or a conclusion that does not yet exist**, including in the future tense, the passive voice, or as an "anticipated finding". This is the specific and most dangerous failure in conference submission, because the deadline precedes the work and the format invites it. Where a result is not yet available, the correct text says what stage the work has reached (K4 §2.1, §2.5).

3. **Never inflate a modest contribution to fit a competitive call.** Confident phrasing applied to a small study produces an abstract that wins selection and disappoints a room, which is a worse outcome than rejection. Match the register to the evidence (K3 §4).

4. **Never invent a citation, a comparison to prior work, or a claim about what the field currently believes**, in an abstract or on a slide. Abstracts are frequently published and permanently citable, and a fabricated claim in one is as damaging as in an article.

5. **Do not assume a discipline's conference culture.** Whether conferences are archival or exploratory, whether abstracts are peer reviewed or curated, whether posters carry status, whether work in progress is welcome, expected session length, and question-session norms all vary sharply between fields and between regions. Where a norm matters and is unknown, ask, or read the previous programme.

6. **Never assume a presenting style, a level of confidence, or a personality.** Advice framed as "be more confident" or "be more outgoing" is not actionable and is unhelpful to the people who most need the skill. Give specific, executable actions instead (Step 11).

7. **Do not generate slide or poster content that presents identifiable participant material** without confirming the consent covers public presentation and recording. Conference audiences photograph slides; treat anything shown as published (K5 §2.4).

8. **Where a talk would disclose an embargoed, confidential or pre-publication finding, flag it rather than drafting it.** The decision belongs to the researcher and, where relevant, to their collaborators and funders (K5 §2.8).

## 13. Best-practice principles

- **The abstract's job is to be selected; the talk's job is to be understood; the poster's job is to make someone stop.** Three different jobs, three different designs.
- **Read the previous programme before writing anything.** It answers the track question, the completeness question and the register question at once.
- **The result sentence is the one that must not be squeezed.** Cut background before you cut findings.
- **Submitting unfinished work is normal; promising results you do not have is not.** The formats built for work in progress exist because everyone is in this position.
- **A listener cannot re-read.** This one fact generates every rule about talk design.
- **A talk carries about a quarter of its paper.** Anyone who has not measured this will overestimate.
- **Cut content is not wasted; it becomes your backup slides.** The best answers in a question session come from material that did not fit the talk.
- **Build to eighty per cent of the slot.** The remaining twenty per cent is consumed by reality.
- **Decide your cut slides before you stand up.** A calm decision beforehand beats a panicked one in front of an audience.
- **"I do not know" is a complete and professional answer**, and it is heard as confidence. Improvised defence of a weakness is heard as the opposite.
- **Everything on your slides is public the moment it is shown.** Assume photography.
- **Networking is a list and a follow-up deadline, not a personality trait.** One specific question asked of one person whose work you have read is worth more than an evening of circulating.

## 14. Worked example

Generic fictional scenario, academic (human-computer interaction, with a practitioner audience).

**INPUT**

A doctoral researcher has a study of how visually impaired users navigate public transport information displays. Data collection is complete (18 participants), analysis is roughly half done, and the conference deadline is in nine days with the conference itself five months away. The call offers three formats: full paper with a 20-minute talk, a work-in-progress track with a poster, and a 5-minute lightning session. The tracks include "accessibility", "public infrastructure" and "user studies".

**PROCESS**

*Step 1 and 4.* The previous programme is read. The accessibility track contains work like this, with three papers the study would sit beside; the user studies track contains predominantly method contributions and would be a poor fit despite the obvious surface match. Track chosen on evidence. The programme also reveals that the community expects a complete study for the full paper track, and that its poster session is well attended rather than a consolation.

*The judgement call.* The temptation is the full paper track, which carries more status and would look better on the record. Honest projection says the analysis will be complete in about six weeks, well before the conference but weeks after the submission deadline. Writing a full-paper abstract would require either stating findings the researcher does not yet have, which is prohibited, or writing an abstract so vague that it fails on contribution. The third option is what the work-in-progress track exists for. The decision is the work-in-progress track with a poster, submitted honestly, with the complete analysis presented at the conference. The researcher records the reasoning, since the same judgement recurs every deadline.

*Steps 2 and 3.* The abstract is drafted to structured fields. Background and objective: two sentences on why display navigation is a distinct problem from web accessibility, which is the framing the accessibility track's reviewers will recognise. Methods: 18 participants, the recruitment route, the observation and think-aloud protocol, the setting, and the analytic approach. Results: the honest form. Two findings from the completed half of the analysis are reported as findings; the remaining analysis is described as under way with its completion date, and no anticipated outcome is stated. Conclusion: one sentence on what the completed findings imply, scoped to them. The abstract remains true if the second half of the analysis produces nothing at all.

*Step 7.* Poster design. The title states the finding rather than the topic. One dominant graphic shows the navigation path of a participant against the intended path, which communicates the finding at three metres without reading. Three sections with an obvious reading order. Body text at a size legible at a metre, which forces the word count down to roughly a fifth of the researcher's first draft. The two most detailed method paragraphs are cut and moved to a handout.

*Step 9 and 10.* The thirty-second spoken summary is rehearsed to the point of being automatic, since it will be delivered perhaps forty times. Ten questions prepared. The feared question is "you have 18 participants, all recruited through one organisation, so how do you know this is not an artefact of that group". The prepared answer concedes the sampling limitation precisely, states what the claim is scoped to, and names the follow-up study that would settle it. Rehearsing it removes the fear, which was the main reason it was feared.

*Step 11.* Six names identified from the reference list, four of whom are attending. One specific question prepared for each. Two sessions chosen for reasons other than obligation. Follow-up commitment: three messages within a week of returning.

**OUTPUT**

An accepted work-in-progress submission with an abstract that was accurate on the day it was written and remained accurate at the conference, a poster whose finding is legible from the aisle, a rehearsed summary, ten prepared answers including the one that mattered, and three follow-up conversations that came from questions prepared in advance rather than from circulating at a reception.

`RESEARCHER REVIEW RECOMMENDED` on the format decision, which trades record-value against honesty and is a career judgement (K5 §2.1).

## 15. Advanced usage

**Using a conference deliberately as a stage in the paper's development.** Presenting before submitting is one of the cheapest forms of peer review available: the question session reliably surfaces the objections reviewers will make, months earlier and at no cost. Record the questions asked (or have a colleague record them), and treat the three that recurred as a pre-review of the manuscript. This works with **15.24 Journal Article Development**, Step 11, and with **15.19 Examiner-Proofing and Defence Strategy**.

**Presenting the same work to two audiences.** A study presented to a methodological audience and to a practitioner audience is two talks, not one talk with a changed introduction. The message, the supporting points, the evidence emphasised and the level of method detail all change. Design each from Step 5 independently; adapting a deck is what produces the talk that lands with neither audience.

**Symposia, panels and organised sessions.** Where you contribute to a session with a shared theme, the design constraint includes the other contributions: the talk must do its job and fit the session's arc, which usually means cutting more than a standalone talk would and coordinating explicitly with the other speakers on what each will cover. The organiser's introduction can carry shared background, which frees several minutes across the session if it is planned.

**Where the standard approach does not fit.** Some fields present work by reading a written paper aloud, which changes the design entirely: the constraint becomes the written text's readability aloud, and the visual material may be minimal or absent. Some communities treat the poster session as the primary venue, with talks as an adjunct. Online and hybrid formats change attention spans, question dynamics and the value of the room, and generally require shorter segments and explicit invitations to interact. Establish the format's actual norms from the previous programme rather than assuming the in-person, slide-based talk is the default.

## 16. Skill chain

**Recommended previous skills:**
- **15.24 Journal Article Development.** Hands over the contribution claim, which becomes the abstract's contribution sentence and the talk's one message.
- **15.10 Data Analysis Chapter Development.** Hands over the results in a state where what is complete and what is not can be stated precisely, which Step 3 depends on.

**Recommended next skills:**
- **15.24 Journal Article Development.** Takes the questions the audience asked and uses them as a pre-review of the manuscript.
- **15.25 Journal and Venue Selection**, where the conference is a step toward a journal rather than the terminal venue, and where the prior conference version must be disclosed at submission.

**Runs well alongside:**
- **15.26 Peer Review Response**, which shares the discipline of answering a hard question without capitulating or bristling.
- **12.03 Research Report Compilation**, for the different problem of presenting to a decision-making audience.
- **13.05 Research Ethics and Consent Design**, for whether participant material may be shown publicly.

---
A Yazi Supplied Skill and resource.