It's a real timesaver for Extract, Transform, and Load (ETL) applications, such as loading a data warehouse. Rows that don't already exist in the data warehouse are inserted, and rows that do exist are updated.
When the MERGE statement was introduced, both an UPDATE and an INSERT clause were required, and the order was fixed (first the UPDATE, then the INSERT). If you only wanted to do one or the other, you would use the existing INSERT or UPDATE statement instead of MERGE. Deletions were always done separately via the DELETE statement.