Macy’s cloud-data program identified Data as a Service as an early Teradata/Hadoop-to-BigQuery migration priority, with forecasting, replenishment, purchase orders, inventory, and other supply-chain domains included in the broader roadmap.
What made it hard
Legacy analytical systems carried years of embedded assumptions. A technically correct copy could still change business meaning, duplicate data, increase operational risk, or expose inconsistent answers across teams.
How I approached it
My work focused on migration patterns that treated source analysis as a first-class step: identifying keys, partitions, and change indicators; building controlled ingestion; validating completeness and restart behavior; and pairing delivery with governed access, monitoring, and reconciliation.
Selected impact
The effort helped establish a repeatable path for moving analytical workloads toward BigQuery while protecting operational systems from heavy analytical use. Public-facing claims are intentionally limited: no unverified cost saving, data-volume, performance, or migration-completion figures are presented here.
Lesson I still use
A cloud migration is credible only when the new platform preserves the controls that make its answers understandable and challengeable.