The current data schema is a bit too normalized - there are 34 tables for some fairly standard objects in Affinity. Because of this, there ends up being many tables that are essentially just rows of ids that serve as links to other tables. This creates many more rows than necessary - a less normalized build out would have less atomic data but aggregated better, meaning that certain objects in Affinity don't result in a high n number of rows for any given create, update, or delete. This not only makes the connector less complex to navigate the join structure but also cheaper given that fewer MAR are needed to achieve the same end result. Additionally, it would be great to select only certain objects to replicate data from Affinity. For example, I'm uninterested in the following objects: file, relationship_strength, and any of the interaction_chat objects. Unselecting these objects would not only reduce MAR and Fivetran costs for users but also reduce cloud storage costs too. Not everyone wants to bring over every single object.