DWH Ingest ALL to one table or Split out to smaller tables for Transformation (ETL prep)

Another DWH Best Practice question our DBs in MongoDB follow a structure as so: *company (DB name) -- collectionId -- collectionId_metrics -- collectionId_dashboards *activity_log (DB name) --collectionId -- collectionId_actions -- collectionId_notification what is best practice on ingestion? Should I extract each 'sub-type' to a separate raw table, or just dump the whole DB into one big table and then parse out upon transformation? ie. TABLE = company or TABLE = activity_log and then transform and split out OR TABLE = company, TABLE = company_metrics, TABLE = company_dashboards, etc?