It is good to hear you are thinking in this direction.
A connector to a EM dset will solve this problem. It will allow users to Pull data into EM instead of Pushing it to Power BI. I mentioned that Excel and Power BI share the “Get data” functionality, so one connector will allow users to pull data into either.
Pushing data into SQL is probably the more elegant solution for advanced solutions, but it adds a layer of complexity that prevents 99% of non-coders to attempt it.
Most of my work just calls for a simple place to store data in one or two tables for import into Excel. The tables being overwritten each time (no complexity of upserting, maintaining a database.)
Writing to a csv file is such a simple solution - except that it is slow and large. Hence my suggestion to use dset files, knowing it would be smaller and hoping it would be faster.
Power Bi really makes it simple to create a connector. Would not take you more than a few hours to figure out. Google “create custom connector power BI”