How to present a research paper at a conference

How long is a conference paper presentation, and how many slides do you need?

Each conference fixes its own timing in the presenter instructions. Published IEEE guidelines show the usual range. Back in 2014, at the IEEE Nuclear Science Symposium, one slot was 15 minutes, out of which roughly 12 were for presenting and 3 for questions. In IEEE MMSP 2021 and IEEE IST 2025, speakers had 10 minutes to talk and 5 for questions. For the 2026 IEEE EIC and IPMHVC conference, contributed papers get 15 minutes and another 5 for questions. EIC has even written that session chairs will strictly enforce the times.

• Plan for the talk time, not the slot. Mostly a “15-minute slot” is 12 minutes of talking and 3 of questions, or sometimes 10 and 5. Check the presenter instructions, and if nothing is written there, ask the organisers.
• Budget about one slide per minute. Some figure you may need to talk about longer, and that time comes from the other slides. For a twelve-minute talk, take twelve slides as the upper limit.
• Put the time on your slide notes also. Mark where you should reach by 4, 8 and 10 minutes. Suppose at 8 minutes you are still behind, then go directly to the comparison slide.
• Build a skip slide. Fix beforehand which result you will drop if needed. Sessions many times run late, and the chair may tell you to finish in 8 minutes only.
• Finish early rather than late. If you go over, the chair will spend your last minute stopping you. That time also comes out of questions, and questions are the main reason you came.

What this looks like at an Indian conference

A lot of conferences held in Indian engineering colleges and universities publish through IEEE as technically co-sponsored events. Others go into a Springer proceedings series. Sessions mostly run as parallel tracks with many papers each, and the chair cannot run late because the next session is waiting. Many chairs ring a bell or show a card two or three minutes before your time ends. Most will stop you if you cross the limit. In the break before your session, give your file to the chair and open it on the hall computer itself to check nothing is broken.

Presenting is usually compulsory. As per the policy of the IEEE Sensor Array and Multichannel Signal Processing workshop (SAM 2026), an accepted paper not presented by an author or qualified proxy will not be distributed as part of the proceedings. Read your own conference’s no-show policy before registering, because an unpresented paper may be dropped from the proceedings.

What should each slide contain in a 10–15 minute talk?

This structure fits most empirical and engineering papers. Theory paper people can put the main result in the results slots, with a rough sketch of why it holds. For a qualitative study, put two or three themes in that place, and one quotation for every theme.

Two habits help a lot. First is the slide title. Most scholars simply write “Results” there. Write the point there as a full sentence, like “Our method cuts training time by 40% at the same accuracy”. Someone reading only your titles should still get the whole argument. Second habit is to keep background short. People in a specialist session already know their field’s history. They came to hear about the gap.

A pre-submission or thesis-level talk before your committee works differently. It runs longer and moves through the objectives one after another. Our guide to the pre-submission seminar explains that structure.

What should you cut from the paper when you make the slides?

You will have to cut a lot, and it will feel like too much. An eight-page paper holds far more than you can speak in twelve minutes. Cut in the order given here.

1. The literature review. Keep only the three to five studies nearest to your work.
2. Equations and derivations. Mostly a method has one defining equation. Show that one, and say in simple words what it does. For all other maths, tell the audience it is in the paper.
3. Secondary results. Each table in the paper does not need a slide of its own. Keep results that support the main claim, and add one result where the method performs weaker.
4. Implementation detail. Hyperparameters, software versions and complete sample characteristics can sit in a backup slide after the thank-you slide, ready if somebody asks.
5. Full sentences. If a slide can be read out like a paragraph, it has too much text. Put short phrases on screen. The full sentence you speak.

One thing you should not cut is the limitation slide line. Specialists in the room trust one honest line on what your result cannot show, more than lot of confident claims.

How do you make figures and tables readable from the back row?

Think that the professor you most want to impress is sitting in the last row, and the hall projector is a dim old one. Conference guidelines give actual numbers for this. IEEE IST 2025 wants 36-point titles, main text of 18 points or more, and at most three colours in a single visual. In 2014 the IEEE NSS guidelines recommended fonts of 18 to 20 points.

• Redraw figures for the slide. A paper figure was made for a printed column, so on a big screen the axis labels and lines look too thin. Make them bigger. Take out the gridlines. Where direct labels can work, remove the legend also.
• Turn tables into one message. Retain just the rows and columns that matter, and put your own result in bold or in a colour. A 12-row results table on a slide is useless for someone in the fifth row.
• Use colour that survives projection and colour blindness. Pale yellow and light green mostly vanish on a projector. Red and green side by side are hard to separate for many viewers. Blue with orange is safer.
• Dark text on a light background you should use when the hall lights are kept on. Try your file on a projector if you get the chance.
• Number your slides. Then the person asking a doubt can just say “slide 8”. Otherwise he has to describe “the graph with the blue line”.

Posters need the same rules even more strictly, as explained in our guide to making an academic poster.

How should you rehearse, and what changes online or in a hybrid session?

Rehearsal

Speak through the full talk aloud with a clock running, minimum three times. Do the first run alone, only to find the time. In the second, record yourself and watch it back. For the third, present to your lab group and let them interrupt with tough questions. Almost every first run comes out too long. Most of us then try talking faster, which only makes the talk unclear, so remove a slide or two.

Learn the opening minute nearly by heart, since nerves are worst at the start. After that, no need to by-heart. Know the reason each slide is in the deck. Once you know it, words come by themselves. Many of us speak English as second language, so practise those technical words that trouble your tongue, and go slow. Indian conference audiences come from many states and countries, so a slow clear speaker is easier for them to follow.

Online and hybrid presentations

• Know who controls the slides. This differs. At IEEE MMSP 2021 the remote presenter had to share own screen and move the slides, while a few other conferences ask for a pre-recorded video. Find out your conference’s way a week before itself.
• Test the setup the day before. Camera should be at eye level. Use a wired connection if possible, wear a headset mic and keep the laptop on charging.
• Send a backup. Email a PDF of the slides to the session chair or host. Then even when your net goes off, the host can go on showing your slides.
• Watch the chat. In hybrid sessions the chair many times reads out questions coming from online attendees. Give a small pause after each one, because the hall audience also has to catch what was asked.

How do you handle questions, including hostile or off-topic ones?

In the whole talk, questions help you most. They show what a reviewer or examiner will ask you later. Leave the thank-you slide on screen during the questions.

• Repeat or rephrase the question briefly. You get a few seconds to think, and everyone in the room hears what was asked. This matters most in big halls and hybrid sessions.
• Answer in two or three sentences. Suppose you made a backup slide for exactly that doubt, go to it. If you go long, others waiting to ask lose their chance, and the next speaker loses time.
• If you don’t know, say “I don’t know”. Something like “I haven’t tested that. It’s a good idea for the next step. May I take your email?” is a strong answer.

Hostile questions

Once in a while somebody attacks your whole premise or says the method is wrong. Keep your temper and reply like it is any normal technical doubt. Tell clearly where you agree (“You’re right that the sample is small”) and where you see it differently. Then give your evidence briefly. If it keeps going round in circles, offer to discuss about it after the session. The audience is watching how you respond, and for them that counts more than who wins.

Off-topic and long questions

At times the question is outside your paper. Politely say so, and give a short reply if possible, for example “That is outside this study, but briefly…”. Or someone gives a long speech with no question at all. Thank them, answer the one point you can, and look at the chair. Moving the session on is the chair’s job.

Write the questions in your notebook the same evening, before you forget them. If you later expand the paper for a journal, these questions become an early draft of your reviewer comments.