Start hammer. Throw a bunch of random brushes together. Take to gameday. Huge success.
Personally I barely plan ahead on paper, for better or worse, it just doesn't work for me, there's no sense of scale, no z-axis on a sheet of paper, and if I'm going to take the time to make sure it is to scale and define precisely the height levels, I might as well start hammer and make up a very rough 1st draft, that's easier for me.
What I do however, is draw general map ideas on paper : objective positions, pathes (simply arrows), fighting areas (large squares), while taking into account major visblocking requirements. Before I do that I generally have a vague idea of the theme of the map and a few specific ideas I'd like to put into the map. Then I build it as I go, loosely sticking to the plan I put on paper.
You're never going to hit the spot with your first alpha anyway, there will be a lot of stuff that you didn't think of, or that didn't turn out like you expected, and you will need to make changes, sometimes large changes, so planning too far ahead isn't the best idea IMO.
Now to give you any specific advice... We'd need to know what you tried, why you think it failed, and what you want to do now...
There's also plenty of interesting reads in the tutorials section.
And well... Frankly, if you think your map is not working and you want to give it up and start something new, that's not unreasonable. Sometimes, the only way to truly fix a map is to start over because there is a fundamental flaw in it. There's no receipe for a perfect map, and if there was they would all play the same. You need to experiment and learn for yourself what works and what doesn't. Different players also have different tastes, you're never going to please everyone, it's up to you to decide who you want your map to appeal to : as many players as possible ? competitive players ? terrible players ? yourself and your friends ?