Write your own SQL query vs views vs table imports

[EnrichedUser]

I am trying to optimize my future models. Previously, I have been writing my own queries joining tables, making aggregates, etc. 

I have since learned that I lose out on the built in query folding and the subsequent access to things such as incremental refresh. 

What is the best way to pull SQL data into the model?

Writing your own scripts

Creating Views and pulling in the entire view like a table in PQ

Pulling in entire table (import) and using PQ ETL

Something else

if there are any solid articles written about what the optimizal method is, id love to read about it as well.

Note:

No single table exceeds 10mm rows, but many have 100+ columns

Thank you

[Fowmy · Accepted solution]

@EnrichedUser 

Is always recommended to do all your ETL at the source level and bring in the data to PBI. Creating Views is the best option as per your case. It will enable Query Folding and Incremental Refresh as well.