AWS makes the numbers, a manual export posts them

Revenue recognition and journal logic run in Spark on AWS, at roughly 20.8m orders a month. The output leaves as a Tableau export, lands as CSV on SharePoint, and is uploaded into Dynamics 365 by hand at a 3,500:1 to 5,000:1 compression with no key to trace it back. Getting data back out of D365 runs through the Entity Store, Power BI and another CSV, with a full historic refresh to 2018 on every run. Pigment ends up fed from two routes that nobody reconciles. Hover a node to highlight its connections; click for detail.