While some may feel that grant proposals should be perfect, I think they should be good enough to be funded and no better. Only three or so people will read the thing, and they’re not allowed to talk about it; you should be devoting yourself as much as possible to the actual work rather than the grant writing.
Having read many grants (some good, many bad), I’ve formed quite strong opinions about what constitutes a good grant.
The ideas are most important
Perhaps this is obvious, but in many grants, the real ideas are not apparent. Don’t hide them (thinking the reviewers might steal them), and don’t just describe an important problem. (We can all mostly agree on what problems are important; the question is: can you solve any of them?) Focus more on what you will do than on what has been done.
If you don’t have a very good idea, it might be best to wait until you do.
Understand the process
It is useful to know what happens to your grant once it is submitted: Who will read it? What are they looking for?
Talk to senior colleagues who are familiar with the funding agency. Enroll in a grant writing course. Read the material provided by the funding agency (such as on the web).
Write your proposal early enough that you can get feedback from colleagues and still have time to make use of their comments.
Reviewers are tired and scatterbrained and not necessarily experts in your sub-discipline
You may wish that the reviewers would spend a quiet day devoted to reading your grant front-to-back, but that won’t happen. They may have ten grants to read and are still trying to do their real work. They’re likely to read your grant in bits at a time, out of order, maybe on a plane. And they’re often old and wear glasses.
Include plenty of white space
Use large fonts
Use simple, blunt text (simple sentences, simple words)
Ensure that the organization of the proposal is absolutely clear
Repeat text verbatim (e.g., of the aims) so there’s no question what you’re talking about
Write their review for them
If the reviewers can borrow your text to explain how the work is important and innovative and why you are the right person (or team) in the right place to do it, they’ll be happy and thus you’ll be happy.
Include nice illustrations
Illustrations can be a lot of work but can bring clarity to complicated things. But they can be a waste of space. Include some meaningful ones and omit those empty of content.
Pay attention to details
Avoid typos, misspellings, missing words, messed up figure numbers, missing letters of support, incomplete or out-of-date biosketches (especially those from collaborators), and inaccurate or missing references.
If the reviewers are annoyed, confused, or sense sloppiness, they won’t write a good review (and some of this may not be conscious on their part).