The pattern we found
A growing SaaS had let architectural decisions made two years earlier carry forward unchallenged. Compute was oversized, storage tiering was non-existent, and unattributed cost lived in shared accounts.
There was no FinOps owner - and without one, every team optimized for their own velocity, not the bill.
The work
We started with a 30-day audit: usage analysis, rightsizing recommendations, reserved capacity planning, and a tagging policy that finally made cost attribution possible.
Implementation ran in phases with monitoring on every change. No optimization shipped without an automatic rollback path.
The discipline we left behind
A monthly FinOps cadence kept the savings from regressing. Twelve months in, the 40% reduction was still in place - and the engineering team had a shared language for cost trade-offs.

