The Prof Hacker offers an interesting post on “The Conference Abstract.”
Some of these points I thought were obvious:
follow length requirements
but if Prof Hacker is mentioning them, perhaps they are not.
Proofreading, an excellent recommendation, requires that you be prepared to send your abstracts in early.
You need to have at least 24 hours to go back over what you wrote and decide if it will work.
Also, I have learned this from personal experience, if you can only give one talk at a conference, only submit one abstract. Make it your best. It’s not considered kosher to send multiples out, where you might be accepted for more than are allowed.