But how do you collect all queries on a table to decide which field is selected the most and interesting for a clustered index? Through long lasting Profiler traces? Or ist there a more efficient (maybe also quicker) way? Are there any DMV's for that?

Found the answer myself in Gail's blogpost! Thanks Gail!

But how do i use SQL Trace or Extended Events to capture my queries? Does anyone of you know step-by-step explanations or existing solutions?