By "mechanism" do you mean "common practice?" In that case, I don't think there is one.

The fillfactor specifies how much data to put the leaf-level nodes of an index while the index is being created. Specifying a value of 50 would fill the leaf-level nodes 50%. The idea is to delay page-splitting. Page-splitting occurs when the node is full but SQL Server needs to place a key on the page. In this case, SQL Server will create a new page, move 50% of the data on the full node to the new page and then place the new key on the proper page. So, the setting you use of the fill-factor might depend on the volitility of your data. That is, how often changes are made.

Now, the downside is that all that empty space causes the index to be bigger, page-count wise. This means that SQL Server has to do more work to read the same number of index keys. Like anything else, it's a balance.

I'm betting that most people either don't play with it or experiment.

-Mike