I've just finished reviewing another set of proposals. Some of the mistakes I've seen are simple things. Often they are things that someone else can catch for you.
- If one critique was your productivity ("it reduces enthusiasm for the proposal in raising concerns about the productivity of the PI"), and you claim in the intro that you have X new pubs and Y new things submitted, make sure that you biosketch and google scholar/pubmed link (the one that the instructions request) is up to date with these pubs. Otherwise it could look like you are stretching things.
- In general don't stretch things. I don't have to say that its cheating to out & out lie about results, but if the reviewers think you're pushing the limits of interpretability, you are sunk.
- Typos. Bad English. Particularly ones that spell-checkers don't get. Sometimes they are worth a smile from a reviewer. More often it says: I was in such a hurry, and I am so disorganized that I did not give the proposal one last read through. One such error might be acceptable. Two raises eyebrows. Three and you've sunk your proposal.
- Ignorance of the literature. This is Scylla & Charybdis. You can waste half a life-time reading other things. You can waste half a life-time because you didn't read other things. It is hard to know who will review your grants (though you look for the roster). But you can make some educated guesses (go check your key words at RePorter - go see who is funded in your area).