Usually the first chance to give a talk at a big conference can occur when you're a postdoc (or maybe even when you're a PhD student if you're lucky). Often it means you get a short slot (15-20 minutes), and depending on the amount of parallel sessions the audience can be relatively small. However, making a big impression on the right people (those who are involved in the organization of other conferences), can give you lots of other opportunities to talk at meetings and meet more senior researchers at speaker's dinners etc.
The first thing to ask yourself when the opportunity arises is "Am I really good enough at giving talks?", and no matter what your answer is, there is always room for improvement. And it does not have to be that much work. Before my first big talk I went and bought a book on giving talks and slide design: Presentation Zen by Garr Reynolds, and it was a big eye-opener for me. I read the book, tried to apply the most important lessons I learned from it, and my first talk was a big success (and afterwards I went and bought more books on the subject). This and similar books are not something you should pick up a week before your talk, but something you need to start reading 1-2 months ahead of your talk.
The major things to take home, if you don't feel that you have time to read the book (which you should) are
- The audience will at most remember three things from your talk, decide what these are
- Every slide should support one of these three things, otherwise remove them.
- You got two minutes to hook the audience, or you will be forgotten, use these two minutes to catch them.
- Slide design should be clean, simple, and clear.
- Complex slides should be built up on the screen, not be shown all at once.
- Bullet points are a big no-no, and text on slides should be minimal. The audience are there to see and hear you, not read your slides.
Blogs and related links on design, talks and posters that might be interesting to you:
