There is a helpful and challenging article The Physics of Physics Colloquia by James Kakalios on The Back Page of the APS News. It is based around old notes Suggestions for giving talks by Robert Geroch.
Both Kakalios and Geroch are worth reading in full, but here are a few random things that stood out to me. [Things I need to keep working on].
"What is the key take-away point that you want to impress on everyone when they leave your talk?"
Divide the talk up, centred around 3 or 4 key messages.
"Figures are easier to understand than words."
"You have been staring at these data and plots for years, but many in the audience have not."
Don't include more than five non-trivial equations.
"It is almost always a disaster to run over time".
Much of this may seem "common sense". However, as management guru Steven Covey said, "Common sense is not common practise." Preparing and giving a good talk requires a lot of discipline, particularly with regard to cutting out material.