If you need customizable reports (filtering and aggregating by any dimension), you cannot avoid scan over non-aggregated data. This is where you cannot optimize

If you need customizable reports (filtering and aggregating by any dimension), you cannot avoid scan over non-aggregated data. This is where you cannot optimize Postgres queries. Postgres becomes slow starting from just single million records.