Tips and Tricks for Giving Talks

``Present to inform, not to impress; if you inform, you will impress.''
--- Fred Brooks

I don't attempt to be comprehensive here, but I do give a few tips and tricks to supplement material pointed to from my page for students.

Tips for preparing your talk

Your slides
The most useful trick I've developed on my own is to use three-by-five index cards to prepare my talks. Generally, what I can write comfortably on an index card fits comfortably on a slide (although in both cases I can pack too much). I perceive that the big advantages are these:

Say early, on the first slide if possible, what is the problem you are trying to solve.

Don't waste time on an outline at the beginning of your talk. They already know it (thanks David Wakeling):

  1. Motivation
  2. Describe the Problem
  3. Outline the solution
  4. Implementation details
  5. Benchmark figures
  6. Ponder the results
It may be appropriate to give a short outline after your introduction and motivation, to tell your audience what tiny fraction of your subject they can expect to hear about in the rest of your talk.

USE LARGE FONTS. Your slides should be easily readable even in a large room. If you can try them out in the room you will use, good. Otherwise, make sure a printout on US Letter paper can easily be read from six feet away.

USE BOLD, SANS-SERIF FONTS. Most fonts are designed for readability at 10-point size. When blown up to large size and viewed in a large room, these fonts are just too thin. Bold, sans-serif fonts are much more readable on slides.

You want your audience's attention on you, not on your slides. Every time you put up a new slide, you lose them while they scan the new slide. So scrub as much ink as possible off your slides.

Use landscape mode. It reduces your temptation to cram too much onto the slide, and it makes it easier to make points on one line.

I get substantial leverage from using the same tool to prepare both talks and papers. For me, that tool is latex, because I am usually presenting either mathematics or a literate program, and latex is the tool of choice in these arenas. The latex package in ~nr/lib/latex/nrslides.sty will help you follow some of these guidelines.

Another trick I use for just about every talk is that I give my first practice talk in an empty room. This trick enables me to smooth out my delivery, revise my slides, and so on, without wasting anyone else's time.

Once you are relatively happy, try to give a practice talk for someone else. Use your practice listeners carefully---each one can hear your talk for the first time at most once. If you're giving an important talk (your first conference talk, your job talk), you should probably plan on giving practice talks to several different buddies. Then you can take your talk to your research group and your advisor for final polishing.

One of the most important things you can do is be sure you finish on time and say what you wanted to say. To run out of time without having presented your best results is the most embarrassing of rookie mistakes. I see it happen occasionally at 25-minute conference talks, and sometimes also at students' qualifying exams. Here are some things you can do to avoid this mistake.

  • Practice your talk repeatedly. One or two practice talks in an empty room will tell you how much of your material you have time to present.
  • If you have never given a conference talk before, you may want to consider writing out a script of every word you intend to say with each slide. My dissertation adviser made me do this for my first conference talk. It was a lot of work, but worth it---the talk was very successful. After years of experience, I seldom do this any more, but I may still script parts of a talk, particularly if the stakes are high (e.g., I am presenting important work to an audience whose opinion matters to me). To reduce the amount of time it takes to prepare the script, I usually
  • Give the talk naturally and spontaneously, not worrying about length, while having a recorder running. I like using the computer to record my talk because I can very easily stop and start playback, as well as seeing the time of my words to the minute and second.
  • Transcribe the recording into notes for each slide.
  • Give the talk again using the notes.
  • Repeat until convergence.
  • If I have audio that is good enough, I may also put it on my web site, so that anyone with web access can hear the talk.
  • Before you give your talk, know to the minute how much time you expect to take. A colloquium I gave in 1999 took 48 minutes; a conference talk I gave in 2001 took 27 minutes. Although my conference presentation was supposed to take only 25 minutes (with 5 minutes for questions), my session chair was willing to let me talk for 27 minutes when I asked in advance. He knew I would not take 28.
  • The beauty of a 25-minute talk is that you can practice it repeatedly. If you are giving a high-stakes talk, it is worth it.

    Tips for presenting your talk

    Back to Norman Ramsey's page for students or Norman Ramsey's home page.