In my experience (I know nothing about this) if I needed to join a lot, I’d actually create new tables which routinely aggregated that data into tables I could query more efficiently instead. Especially if it was analytical data which was used for things like dashboard reporting.

It seems excessive at first but the performance is so dramatically better that it makes a lot of sense. Especially if the queries are part of your hot path. I always thought of it like having core business logic tables, then utility tables which are essentially derived from those core tables.