The Conference Talk: What Twenty Minutes Can Actually Carry

The commonest failure in conference presentations is not nerves or slides. It is attempting to deliver the paper.

A paper can be read twice, at the reader's pace, with the tables in front of them. A talk happens once, at your pace, to people who have already sat through four others. These are different media, and the material has to be rebuilt rather than compressed. For a concrete operational example of how a workplace tool approaches a related coordination problem, see learn more.

Conventions differ by discipline. Slot lengths, the expected level of technical detail and the balance between motivation and method vary considerably. What follows is general; check what your field expects.

What a talk is actually for

Not to convey the results. The paper does that, and anyone who needs the details will read it.

A talk is for making people want to read the paper, and for making them able to ask you a useful question while you are standing there.

That reframing changes what belongs in it. The full specification, the robustness checks, the complete literature — all of it belongs in the paper and almost none of it belongs in the talk.

One consequence worth stating plainly: if the audience leaves able to state your question and your finding in a sentence each, the talk worked. Everything beyond that is optional.

How much fits

Less than people think, and the arithmetic is unforgiving.

Twenty minutes is roughly 2,000–2,500 spoken words, and a talk needs pauses, so plan for less. That is about the length of a long blog post — not a paper.

One main point. Perhaps two. A talk with five points delivers none of them.

A slide takes longer than it looks. A minute per slide is a common planning figure and it is optimistic for anything with a figure on it.

Leave time. Finishing two minutes early is a gift to the chair and to the audience. Overrunning is the one presentation failure everyone notices.

A structure that works

Not the only one, and a reliable default.

The question, in plain language, in the first minute. Not the literature, not the motivation, not your institution. What you are trying to find out. If people do not know what the talk is about by minute two, the rest is spent recovering.

Why it matters, briefly. Two or three sentences. Whoever is in the room is already in your field; they do not need convincing that the field exists.

What you did, at the level of the idea. The design, the intuition, the identifying variation. Enough that the result is interpretable, not enough to replicate.

The result. With the single most important number or figure shown clearly.

What it means, and what it does not. The limits stated by you rather than discovered by the audience is a sign of a careful researcher and it heads off the obvious question.

Stop. No summary of the summary.

Practical points that matter more than they should

Rehearse aloud, timed, standing. Reading through in your head is roughly 30% faster than speaking, which is how talks overrun. Once out loud is worth ten times reading it silently.

Rehearse the first two minutes until they are automatic. That is when nerves are highest, and having them memorised carries you into the part where you are talking about work you know.

Do not read from the slides. The audience reads faster than you speak, so they finish the slide and then wait. A slide should be what you point at, not what you say. See slides that support a talk.

Say the number aloud when you show it, and say what it means. A table on screen with no verbal interpretation is a table nobody reads.

Have a plan for going over. Know which two slides you skip if the previous speaker overran. Deciding this in advance is the difference between adapting and panicking.

Check the room and the equipment before the session starts, not during your slot.

Presenting work in progress

Common at conferences and frequently done badly, in both directions.

Say clearly that it is in progress, once, early. Not apologetically — as information that shapes what feedback is useful.

Ask for what you want. "I would particularly like reactions to the identification strategy" directs the discussion to where it helps you. Without it you get comments on whatever is easiest to comment on.

Do not present nothing. A talk that is entirely motivation with no findings frustrates the audience and wastes your slot. Show what you have, even if it is preliminary.

See presenting work that is not finished.

The parts people neglect

The transition between sections. A sentence saying where you are going — "so that is the design; now the result" — costs three seconds and keeps the audience oriented.

Signposting the structure once, at the start. Not a table of contents slide; one sentence.

The last slide. It stays up during questions. Make it your main finding or your contact details, not "Thank you" or a references list.

Speaking to the back of the room. Volume and pace matter more than content for the people who cannot hear you.

Handling the slot you were given

A short slot is a constraint on scope, not on quality. A ten-minute talk delivering one point well is better than the same content rushed.

A long slot is harder, not easier. Forty minutes needs genuine structure, and it is where people revert to walking through the paper.

A parallel session with six people in the room deserves the same preparation. Those six are frequently the people most relevant to your work, and one of them may review your paper.

The uncomfortable truths

Most of the audience is not listening for most of the talk. They are thinking about their own presentation, or their email, or the last talk. The structure above exists so that someone tuning back in at minute twelve can still follow.

Nobody will remember the details. They will remember, at best, your question and one finding. Choose which ones deliberately rather than hoping.

The questions are the valuable part. The talk is the price of admission to a conversation with people who know your field. Prepare for the questions at least as carefully as for the talk. See handling questions.

A short checklist

• [ ] The question stated in plain language within the first minute
• [ ] One main point, chosen deliberately
• [ ] Rehearsed aloud, timed, standing
• [ ] First two minutes automatic
• [ ] Two slides identified to skip if time is short
• [ ] The key number said aloud and interpreted, not just displayed
• [ ] Limits stated by you
• [ ] Final slide is the finding or your contact details
• [ ] Finishes two minutes early

For an independent authoritative reference, consult the MIT Communication Lab for academic communication guidance.