For me the second use case is used more often because I’ll typically have an underlying “stats” table that continually grows. If I’m applying groupings to the underlying table, then I would have to ungroup the table to do spot checks of my table (to ensure new data is being added correctly).
However, I think grouping of the base table is easily done with a view off of the underlying table. Coming from Excel, I still prefer creating a summary table when the goal is to create a neat report where I don’t want to show irrelevant data points to the end user. Similar to using a filter to get summary stats, SUMIF() and LOOKUP() can be used (thread here).