Question for #data people -- I'm trying to understand headless BI. In particular, what it solves that SQL doesn't. In other words, why dbt (pre-analytics layer) isn't enough.

From what I can tell, it solves the problem that SQL can't easily be parameterized. You can make views that slice and dice your transactional data, but those views would hardcode a bunch of decisions better left up to the consumer. You can also denormalize the heck out of your data to make all conceivable queries easy, but then you end up with an analytical table that's way too tall and wide.

It seems like what these semantic layer / headless BI tools do is apply the metadata to your transactional data that allow for BI tools to offer slick query builder interfaces, which ultimately are generating SQL. Furthermore, the logic for different types of analysis can be standardized and controlled, compared to people handwriting their queries.

Do I have this right?

#semanticlayer #analyticslayer #headlessbi #businessintelligence #analytics #dataanlytics #datamastodon