It is flexible, modular, reduced maintenance, increased accuracy. You can develop common processes (graphs or subgraphs) to be called by other graphs. You can package them into jobs. You can create user facing data apps that provide a simple user interface/page where the users can attach their file(s), run the process, and receive their output. You can create/maintain common layouts/definitions in one place and use them across all processes.

It has awesome GUI that is filled with lots of capabilities like logging, error handling, validations etc. In addition, it connects with various Softwares like Salesforce, Oracle DB. It has repeatable graphs that can iterate over various implementation phases of a project.

A small, but very important piece is the counts of data on the connection edges.