“At one point, things got so bad that it was faster to dump the entire database into CSV files and use Excel! Maybe this really is the right solution for ad hoc queries, but it just seems dirty and inelegant.”
If you have a really small dataset, Excel seems like a great tool for analyzing your data. Excel is probably even easier to use than SQL for nontechnical users.
For data that can fit in one machine, SQL + dump to an Excel-readable format regularly seems like a pretty good interface.
It sounds like the author didn’t have a particularly sound reason for choosing “NoSQL” in the first place.