EKS Cost Right-Sizing After a Health Check

Client: Series B B2B SaaS (anonymized)
Package: Health Check
Timeline: 10 days
Outcome: ~38% lower monthly EKS/compute spend within one billing cycle after priority fixes

Problem

AWS bill had roughly doubled over two quarters. One senior engineer “owned” the cluster beside product work. Nobody had a prioritized list of what was waste vs what was load growth. Leadership wanted cuts without a multi-month platform rewrite.

Approach

1. Read-only Health Check across security, cost, and reliability
2. Ranked findings: idle capacity, oversized requests, missing HPA/VPA signals, leftover non-prod namespaces on prod-class nodes
3. Handed the team a P0/P1 fix list they could execute without us
4. Optional half-day review after the first right-sizing PR merged

Results

• Monthly compute-related spend down ~38% vs the pre-engagement baseline
• P0 security items (over-broad RBAC, public Service exposure) closed in the same sprint window
• Clear split: DIY for the backlog vs optional Managed Support for upgrades

Takeaway

For SMB teams, a fixed-price Health Check often pays for itself when cost is the burning issue — if the output is a ranked backlog, not a 40-page slide deck.