I work in a development environment where data is relatively small, even when we do data migration jobs.

This gives me the luxury of running DBCC DBREINDEX as part of the nightly maintenance plans.

On a client's site we scheduled in a DBCC DBREINDEX and noticed that a 600Mb database shrunk to 250Mb.

Obviously we could have shrunk the physical file as well but as data is input at quite a high rate we decided to leave the file as is to prevent an unscheduled  resize happening. 

This shows that on very large databases there is the potential to save on backup time as well.