How to manage tables built with many other tables and dependencies.

[vix35]

I want to implement this framework one more time, but in the past I had problems dealing with dependencies and where/how to process logic using multiple tables:

The stage B usually runs on a schedule, processing as many tables as there are in that point in time, usually I used this stage to only enrich tables, did not do more complex logic involving other tables in the dataset because I had no way to know if they were already processed with the data I needed, so I always ended in the post-stage (now I think is called analytic) with the same amount of tables that in the pre-stage, and I created another custom stage where I did all the logic involving multiple tables across datasets, considering there is already a stage called "analytics" I think I am doing something wrong, and wanted to know if you could help me to figure out what should I do when implementing it again.