How we (used to) structure our dbt projects

Thanks for the post @claire . I find it interesting that there is a lot of push in the documents around snapshots, and the use of them ‘directly from sources’, yet you make no mention of them in their outline. Do you not use them? The premise of SCDs is core to the Kimball approach, and we are struggling with see how snapshots meet all of the demands that are placed on SC-‘Dimensions’ - but rather, see snapshots as a good way of getting ‘snapshot’ fact tables.

Can you share your perspective on snapshots, where they should be leveraged in the flow (a well thought out one), and what you do when reality strikes (need to build history to before the day you started snapshotting)?

Thank you,

1 Like

Snapshots don’t live in the models directory of a dbt project, so aren’t really in scope for this article. Have you looked at Building models on top of snapshots and the other articles it links to?

I see your question in the community Slack didn’t get a heap of traction - I want to think a lot more about best practices for snapshots in the new year and will try to fold in some answers to that sort of thing :crossed_fingers: