Speaking at a conference: getting selected and being good

What a call for papers actually asks for

Most conferences fill their programme by inviting people to apply for a speaking slot, usually through a call for papers (CFP), a public form asking for a title, an abstract and often a note on who the talk is for. The organisers use it to build a shortlist before the event, not to test how well you can write, so the proposal has one job: let a reviewer who has never met you decide, in the couple of minutes they spend on it, whether your talk belongs in their programme.

What the reviewer is actually checking

A reviewer working through fifty submissions for twelve slots is looking for three things: fit, specificity and evidence you can deliver. Fit means the talk matches what the track promises. Specificity means the abstract names an actual argument or method. Evidence means something in the proposal, a past talk, a piece of writing, a clear line about your experience, tells the reviewer you can hold a room.

Describing what you will actually say

The abstract should let a stranger finish it knowing what they would leave the room having learned. "An introduction to API design" describes a topic. "Why versioning an API from day one avoids the rewrite most teams do in year two" describes a talk. Write the title and abstract last, after working out the one thing you want the audience to take away, and build the description around that.

The mistakes that get proposals rejected

• A generic abstract sent unchanged to several CFPs, mismatched to the specific track or audience level named in that particular call.
• A proposal that describes a product or service reads to most reviewers as a pitch, which they mark down accordingly.
• No stated takeaway, so the reviewer cannot tell what the audience leaves knowing that they did not know before.
• A title that oversells relative to the abstract, or an abstract that oversells relative to what can credibly be covered in the time slot on offer.

One thing worth checking before you submit: some conferences run what is called a pay-to-speak programme, where a slot is effectively bundled with a sponsorship fee rather than allocated on merit. That is a different kind of programme from one that reviews proposals blind, and it changes what a proposal is actually for. Designing a Conference Programme sets out how to tell the two apart from the organiser's side, and Finding, Choosing and Looking After Speakers covers what organisers weigh up when they read a shortlist. A worked example of a submitted abstract, alongside the changes that took it from rejected to accepted, is in Worked Examples and Templates.

A conference talk holds a room when it does one thing. A single argument, one shift in thinking the audience should have by the time you sit down, one idea worth twenty minutes of somebody's attention. Trying to fit a talk's worth of content and a workshop's worth of nuance into the same slot usually explains why a talk feels rushed and unfocused: the audience cannot tell what they were meant to take away because the speaker never picked one thing.

The shape that carries a room

Most talks that hold attention follow a shape with three visible parts: an opening that states what the talk is about and why it's worth the next twenty minutes, a middle that makes the case with two or three points rather than seven, and a close that leaves the audience with a single sentence they could repeat to someone else afterwards. Listeners can't reread a paragraph the way a reader can, so the structure has to do the work that re-reading would otherwise do: tell them where you are, when you move to the next point, and when you're closing.

• An opening that states the point of the talk in the first minute
• A middle built around two or three points, each with its own short section
• A close that restates the one thing you want remembered

A worked example of this shape, filled in with sample content, is in Worked Examples and Templates.

Timing is arithmetic, not instinct

A talk written to fill the slot exactly rarely fits the slot when delivered, because reading a script in your head is faster than speaking it aloud, and nerves add pauses that don't show up on the page. Read the talk aloud, with a timer, and cut content when it runs long. Check with the organiser whether the advertised slot length already includes questions or sits on top of it, and build in a margin either way.

Slides are a visual aid

A slide should carry a chart, a diagram, a photograph or a single phrase. A dense slide competes with your own voice for the audience's attention and usually wins. Large text, strong contrast between text and background, and one point per slide make a deck easier to follow for everyone in the room. If the organiser has arranged live captioning (CART, where a captioner types what's said onto a screen or feed in real time), send slides in advance so names, acronyms and technical terms don't trip up the transcript. What organisers are expected to provide here is covered in Accessibility and Codes of Conduct.

Rehearsal aloud finds what silent reading doesn't

Reading a talk through in your head tells you whether the argument holds together. Saying it aloud, standing up, tells you where the sentences don't work as speech, where you run out of breath, and where the timing actually lands. Rehearsing in front of another person, or recording yourself and watching it back, surfaces the moments a script alone can't: the point you skip over too fast, the transition that doesn't make sense out loud, the joke that needs a beat of silence to land.

What ready actually looks like

A talk is ready when you can deliver the first two minutes and the last two minutes without looking at a slide or a note, because those are the parts an audience remembers regardless of what happens in the middle. It's ready when you can say what the talk is about in one sentence, and when you know which points survive if the projector fails and which don't. A talk that only works with the deck running perfectly is not ready. A talk you could deliver from a blank room is.

A conference talk does not end when you leave the stage. What you do in the following week decides whether the talk was a one-off or the start of something, and most speakers waste that week by doing nothing at all.

What to share afterwards

Slides only help an audience that was not in the room if the slides can stand on their own. A deck built to support a live talk, with a single word or image per slide, will not make sense to someone reading it cold. If you want the material to travel, write a short version of the talk as a page or a set of notes and share that alongside the slides.

• The slide deck, if it can be understood without you narrating it
• A link to the recording, once the organiser publishes it
• A one-paragraph summary of the argument, for people who will not watch or read the rest
• A way to contact you, if you are open to follow-up questions

Post this within a few days, while the talk is still fresh for anyone who was in the room and might share it on. Waiting three weeks means posting into silence.

Handling questions in the room

A question you cannot answer is not a failure. Say what you do not know, offer to find out, and move on. The failure is guessing an answer and being wrong in front of a room that includes people who know better than you do.

A hostile question is different from a difficult one, and the two need different responses. A difficult question deserves a real attempt at an answer. A hostile one, especially one aimed at you, is a moment to stay level, decline to engage with the tone, and answer the substance if there is any. If a question crosses into harassment, that is not something you are expected to manage alone: the event's code of conduct exists for exactly this, and the organiser should be told, during the session if it is serious enough or immediately after if it is not.

Feedback, and what it is actually telling you

Written feedback from conference attendees is often thin: a score out of five and not much else. Take the pattern seriously and the individual comment lightly. One person saying the pacing was slow is one opinion; five people saying it independently is a structural problem in the talk.

The more useful feedback usually comes from the organiser directly, because they have seen enough talks to know what a good one looks like and enough audiences to know what this one wanted. Ask them plainly, after the event, what worked and what did not. Most will tell you, because a speaker who improves is easier to programme again.

Turning one talk into the next one

A single talk becomes a track record when you can point to it. Follow up with the organiser to ask if a recording exists and whether you can reference the talk in future proposals: a link to a past talk, on the same or a related subject, does more for a proposal than a description of what you plan to say. Reviewers reading a call for papers (CFP) are choosing between people they have not seen speak, and a working link removes the guesswork.

From there, the honest path is to keep submitting to conferences whose audience genuinely overlaps with the talk you have. A speaker who does five relevant talks a year builds a reputation in that field. A speaker who does fifteen unrelated ones mostly builds a long list, and audiences at the next conference will notice the difference within the first five minutes.

For what organisers are actually looking for when they choose speakers, and what they owe you once you are booked, see Finding, choosing and looking after speakers. For the earlier stages of getting to this point at all, the rest of this guide covers proposals, preparation and the day itself: see Speaking at a Conference.