Regional retailer: less back-office overtime, lower cloud bill
Four-store specialty retailer. Purchasing and inventory lived in emails and a shared drive; AWS and GCP resources from a prior vendor were oversized. We fixed the operational gaps first, then rightsized the bill.
Overtime was covering broken process
Store managers emailed PO requests. HQ re-entered them. Stockouts and overstocks were normal. Separately, AWS and GCP VMs ran 24/7 for apps that only mattered during store hours.
Purchase workflow + inventory sync + rightsizing
We put purchase requests into a simple approval flow, scheduled nightly inventory sync with Python jobs into HQ, and shut down or resized cloud resources that did not need full-time capacity. No big-bang transformation — three concrete workstreams with a store pilot first.
Main pieces of the work
Purchase requests
Store → buyer approval with quantity history.
Inventory sync
Python nightly jobs for counts and transfers visible at HQ.
Multi-cloud rightsizing
Schedules and SKU changes on AWS and GCP for non-prod and batch jobs.
Monthly review
Cost and stockout metrics reviewed with ops.
- Python
- AWS
- Google Cloud
- POS / inventory APIs
- About 22% reduction in monthly IT and cloud operating spend
- Fewer emergency POs after the pilot store rolled out company-wide
- HQ sees overnight inventory positions without chasing emails
- Store managers stopped maintaining personal spreadsheets for orders
Paying for overtime and idle cloud?
We look at the process and the bill together on AWS, Azure or GCP.