I'd go one step further and be more scientific, why not use automated A/B testing? Like you make a new revision of a lecture, half the class gets the new one and half the class the old one. Then you run some form of quiz on the material. If you have at least a few hundred students or ideally a few thousand you should pretty quick get statistically significant answers. And you could test with short/medium/long explanations to see whether you're beating down open doors or areas they'd benefit from more detail and examples. In the interest of fairness you should of course make all variations available after they've had the quiz, perhaps also with stats to see what quiz questions their path may not have covered as well as the others.
Perhaps that could even develop into a preference system, everything from "I want to learn the essentials for a passing grade quickly" to "I have learning difficulties, give it to me slow and in detail" to "I want to ace this class, give me in-depth insight". Or some form of branched interactive learning, if you grasp 80% quickly you don't need that in more detail but the 20% you struggle with have a more detailed explanation. I think I'd love a system where my hand was on the throttle, it's going as fast as I want it to go not as fast as the professor thinks it should go for some average or "no child left behind" student. Personally I tend to prefer to read the book simply because there's too much time spend on things I already understand.