I usually dislike talks that list many results without showing any proofs or mathematical ideas; such talks just teach you nothing in the end. I also dislike talks where everything is proved rigorously; such a talk will inevitably contain too many details I don't care about, and moreover it is easy to lose the thread in such a talk, especially if the speaker doesn't take care to give people opportunities to catch up again . I prefer a balanced mixture, where some short proofs and examples are given to help my intuition about the results announced.
A. Georgakopoulos: How to give a talk that is not too bad.