Personally, I like to think of a more global, more easy to pick/understand concept before hand.
Something like: pirates attack a merchandise ship, or bank rubbers try to get to the safe.
Then, I split the main idea into smaller manageable events... a bit like in movies.
These become the different control points/stages of my TF2 map.
Once I've got a basic layout of the whole map, with the different goals/objectives I play test it.
Sometimes people don't get it, sometimes they do. Depends on the layout, the physical representation of your original idea. Sometimes you need to make it more obvious, sometimes it is, but people don't like the way you pulled the whole thing off. Sometimes they want more secrets, sometimes they want more unrealistic events happening... sometimes they ask for more realism.
Level design is like playing around with molding paste, it's fun and flexible. That means, it doesn't matter if you redo your work 70 times, you know that it'll get better every time. You learn and get better everytime you show your work to people around you.
There's no magical formula: presenting your ideas and making iterations based off of what people like and don't is the best way to go.
What's cool about mapping and sharing it on the internet is that we aren't restrained to have our work judged by a handful of people inside a small company, but we get to hear about hundreds of people's opinions, for free, and that's priceless.
Just be open-minded and accept criticism and every iteration of your map will get better until it reaches a good balance between your original idea and what people like when playing the game.
No predefined or better way to create new maps for a game...
--> Just know the game.... get involved and keep an open mind and be flexible.
MQ