Best practice

https://docs.getdbt.com/guides/best-practices

  • don’t create high number of specific job → wasting time to recreate some core table solution: start with incremental job and decide what need to be built more than once a day
    • use tags/ exposure to select those specific needs
    • build everything out from the source if the source update frequently
    • build out particular sub-directory to get all data for specific needs (eg finiance)
    • take advantage of union and intersection to select the only target models