Changes only file needs to be sent to the customer.
I have not actually used the Find Changes shape but I had considered it for some of my processes at some point.
This shape will detect changes in a document that goes through it. It does store the last document that ran through it in the file system on your atom. I believe it stores a different document per unique key value supplied which is why I opted not to use it for my processes. My use case it more event driven and I have potentially thousands of unique IDs and I did not want all of those files persisted on my atom.
You can read more about it here: Find Changesshape
The Find Changes shape is most useful for large data sets or mainframe files where a full capture of the data always needs to be retrieved from the source system. An alternate process design should be considered if you can simply retrieve the modified data per process execution because it is naturally more efficient.
Without know more about your use case that is about all the information I can provide.
Retrieving data ...