It does make sense. Very few companies with have ALL necessary data curated and enhanced with all required and potential analytic attributes for identifying segments. Moreover, the speed at which attributes can be created in query service is likely faster than the technical investment to do at the enterprise source level (Discover, Requirements, Build, Test, Deploy). This is especially true for attributes or segments for testing that may be removed or revised if they do not work as planned. No point in paying teams of people to integrate and create attributes that fail to generate value and need to be tossed. Once winning attributes are discovered, that is the right time to push their requirements upstream.
I have also used the Query Service to flatten data sets from dimensional models to avoid the number of joins in my profile merge rules. Could this be done outside AEP? Certainly if people, tools, and skills are also available to do the work given project time constraints. In my example, it would have taken a couple weeks and at least that many development hours with another technical team to have them implement the data flattening. I did it with Query Service in an afternoon.
It can get more complex if you are also using the Data Science Workspace (DSW). Much of work done in Query service can be done in DSW. DSW is a bit more tedious to set up analytic services that provide similar functionality to scheduling queries and required skills are more than SQL.
Best Regards,
David
