Maintainable ETL Practices at Stitchfix
While our Algorithms group does have a large, amazing team of data platform engineers, they do not and will not write ETLs to support the work of data scientists. If we break this large, single query into multiple queries and write the results of each stage separately, we make this pipeline more maintainable. This last example leads to one of the greatest benefits of splitting this query into separate stages: We can reuse our queries and our data to support different use cases.
Source: multithreaded.stitchfix.com