If I'm presenting my research (which is what I do almost every time I present), then I need to do two things: teach the audience about what I did, and show it was worth it with performance results.
As dhimes points out, teaching is different than persuading. If I'm going to teach something, I need figures. Words alone are not enough. In high school and college, my teachers would fill black-boards with figures and equations while they explained things. The first thing my intro to physics professors would always do is draw a picture. That's what I'm doing, except I don't draw my figures one-the-fly as they did. Think about when you first learned the various sorting algorithms. Would you have understood them without watching someone work through examples, moving numbers around as the algorithm progressed? This is very much the same thing.
As for performance results, I need to show multiple graphs that test various aspects of the system to give the audience the impression that everything works as advertised. (Just an impression is enough - they'll read the full paper if they're interested in being thoroughly convinced.) Just showing the graphs isn't enough, of course. I always take the time to explain what the experiment was in the first place, what the axes are, what each plotted line means, and which direction is "good" (for example, if it's scale-up, high is "good"). Finally, in case none of that sinks in, I explicitly tell them, "What we learned from this graph is: ___."
These things are probably going to require more than three slides.
As dhimes points out, teaching is different than persuading. If I'm going to teach something, I need figures. Words alone are not enough. In high school and college, my teachers would fill black-boards with figures and equations while they explained things. The first thing my intro to physics professors would always do is draw a picture. That's what I'm doing, except I don't draw my figures one-the-fly as they did. Think about when you first learned the various sorting algorithms. Would you have understood them without watching someone work through examples, moving numbers around as the algorithm progressed? This is very much the same thing.
As for performance results, I need to show multiple graphs that test various aspects of the system to give the audience the impression that everything works as advertised. (Just an impression is enough - they'll read the full paper if they're interested in being thoroughly convinced.) Just showing the graphs isn't enough, of course. I always take the time to explain what the experiment was in the first place, what the axes are, what each plotted line means, and which direction is "good" (for example, if it's scale-up, high is "good"). Finally, in case none of that sinks in, I explicitly tell them, "What we learned from this graph is: ___."
These things are probably going to require more than three slides.