Do's and don'ts for talks

Jan Vandenbrande

This is a collection of notes of things to do or avoid for your presentation. It's loosely based on many of my own past mistakes and similar to Jim Blinn's talk at SIGGRAPH. The main aim is to present your work as best as possible, to invoke interest and have your audience follow up by reading your paper.

Tell us what problem you are solving and why we should care

It may seem obvious to you what the problem is, and no doubt, you care for your work a lot, but many of us may not be actively involved with your area of study and need this introduction. If possible, provide an intuitive explanation, without jargon or requiring familiarity with other works, of the problem that motivated your work. If we don't understand the problem, we certainly will not understand the significance of your solution.

Think of the following. If you had 30 seconds to tell your dean, or your future employer, about the problem you are solving and why it's important, what would you say? It's called the elevator speech and it's also very useful as an opening statement for a discussion in the hall ways.

Focus on the main points reported in the paper

Your task is not to cover the entire paper at the same level of detail but to give your audience a general idea of what you did and to invoke enough interest for them to read your paper. This means that you should focus on the main points of the paper. After all, do you really think that your audience will remember all the small nuances of your work after listening to 30 other presentations in a jet lagged induced slumber? Providing intuitive insight is often more valuable than the minutia of the approach.

Skip the outline or overview

Here is the outline of 90% of all talks:

  • Introduction
  • Relevant literature
  • Approach
  • Conclusions
  • Future Work
  • Acknowledgments
Do we really need to see this? Is it worth 30 seconds of your precious talk time? Usually not, so skip it unless there is a very strong reason for it. For example, if you are showing 3 different approaches, you may want to make sure the audience does not get lost.

Skip the literature review

The literature review should be in the paper, but in the talk, we just want to 
hear what you have done. Compare your work to others only when 
presenting your results, and say how they are better.


Keep your backgrounds simple

Busy backgrounds are an enormous distraction to your presentation. They add little or no value to getting your point across, so don't do it. Do choose colors for enough contrast between the background and the text to be legible. Blue and yellow contrast well because our eyes are least contrast sensitive for blue and most for yellow. White and black always works. Magenta and Cyan on the other hand are bad contrasting colors. Keep in mind that some of us are color blind and cannot distinguish certain color combination.

Special effects

Use animation or other special effects sparingly and only if it helps explain the point. Otherwise it's just annoying and adds nothing to your talk.


Use pictures, some text

Some people are visually oriented, some are textual. Ideally, you need to cater to both. Keep the pictures simple and the text to a minimum. Do you really think your audience is capable of listening to you, looking at the picture and reading the text all at the same time? Make the text large enough for everyone to see. If you print out your slides and cannot read the text at arms length, then your text size is too small. For complex equations you may need to chop it up into smaller bits so it's still legible from the rear of the room.


Don't cram too much into one slide

It's easier to focus on one item per slide than to focus on several items. Cramming more into one slide will not make it any clearer nor will your talk go any faster. Count on 30 seconds to 2 minutes per slide, which means that for a 20 minute talk you should have fewer than 20 slides on average.

Don't read talks verbatim

You are the expert of your work. Therefore you should have no problem talking about it. Your slides should capture the main points of your talk, which you can use as a reminder to yourself on what to say. If there are points that are important which are not on the slide, ask yourself why they are not there, and if there are items on the slides which you don't mention, consider removing them. The slides and your talk should be well coordinated. This also helps those in the audience who have difficulty understanding you and revert to reading the text. If what you say is different from what is on the slides, ask yourself why and correct it.

Whatever you do, don't read or memorize your talk because it will make for a very stilted presentation. We all have different techniques that work best. For example, just remember the main points you want to bring across and present it like a story you are telling your friends. Don't get side tracked into minor details. Alternatively, explain the pictures on each slide. The only exception is if you have difficulty speaking English, in which case it may be better to read out the talk