Good overview!

For ERP and similar systems data cleaning should be performed as much as possible in the source, otherwise quality acceptance for the various data products will become a nightmare. One can develop data quality assessment reports on the raw data from the lakehouse. It will also help to validate that data synchronization works as expected.

Data enhancement and consolidation are the areas that it makes sense to address in the lakehouse, though for data enhancement one should move as much as possible the data into the source systems.

How many strategic decisions have you seen taken based on data lately? How many poor decisions? How big must be the variation in data for such a decision to be completely wrong?