$43k off
their monthly bill.
In 30 days.
A SaaS platform serving high-volume weather data to enterprise customers had watched their AWS bill cross seven figures annually for the first time. They knew there was waste — they just didn't have a free engineer to chase it down.
Outcome
Before
$173,402
monthly AWS spend
After Koali
$130,189
monthly AWS spend
Annualized savings
Realized, ongoing
$518,556
$173k
Starting monthly spend
$130k
Ending monthly spend
30 days
End-to-end engagement
$518k
Annualized savings
the challenge
A bill growing faster than the team could chase it.
The client's AWS spend had ballooned over the previous year as their customer base grew. They'd crossed the $1M/year threshold for the first time, which triggered both an internal finance conversation and a new opportunity: spend at that level qualifies for an AWS Enterprise Discount Program — but they'd never negotiated one.
Internally, the engineering team knew there were savings on the table. EC2 was over-provisioned. Dev and QA environments ran 24/7. S3 had years of accumulated archive data sitting in the wrong tier. There was no Savings Plan or RI coverage. But every engineer was heads-down on the product roadmap, and infrastructure work that didn't ship a feature couldn't get prioritized.
The constraint that made the engagement most interesting: deployments were only allowed overnight on weeknights, and they weren't an infrastructure-as-code shop. Every change had to be made by hand, in a tight window, without breaking a production data platform that customers were actively querying.
the audit
Four days of read-only digging.
We connected with a read-only IAM role, pulled a year of Cost Explorer data, and ran the full optimization checklist. By day four we had a complete inventory of waste with dollar impact on every line.
Oversized EC2 fleet
Production workloads were running on instance types two sizes larger than their actual utilization warranted. Weekly peak usage rarely crossed 40% on the largest fleet.
Zero commitment coverage
No Savings Plans, no Reserved Instances, no Compute Savings Plans. Every compute hour was being billed at on-demand rates despite years of stable baseline usage.
S3 lifecycle gone stale
Archive data from years prior was still sitting in S3 Standard. Lifecycle policies hadn't been touched since the early days. Multiple terabytes of data that should've been in Glacier or deleted entirely.
Always-on dev and QA
Non-production environments ran 24/7, including weekends and holidays. The team had never built the automation to spin them down outside business hours.
No EDP in place
Now over the $1M annual threshold, they qualified for an Enterprise Discount Program with AWS — but had never engaged the procurement conversation.
Untracked data transfer
Cross-AZ traffic between data services was a meaningful and growing line item. Nobody had ever investigated the routing patterns.
the execution
Two weeks of overnight work.
The plan was approved on day six. Execution had to happen in tight overnight weekday windows on a non-IaC environment, so every change was carefully sequenced with manual rollback paths documented in advance.
Week 1
Right-sizing & lifecycle work
Resized the production EC2 fleet to match actual utilization. Pushed updated S3 lifecycle policies to transition cold data to Glacier and delete archive data the team confirmed was no longer needed. Stood up basic scheduling for dev and QA environments to power down nights and weekends.
Week 2
Commitment coverage & negotiations
Layered Compute Savings Plans across the now right-sized fleet at appropriate term and payment options. Opened the EDP conversation with AWS account management on the client's behalf and structured a multi-year commitment that landed an additional discount on top of the technical savings.
Day 30
Validation & handoff
Verified realized savings against the Cost Explorer baseline. Documented every change made, including rollback steps, so the client's team could maintain the new state. Final invoice followed only after savings showed up on their bill.
the result
A quarter of the bill, gone.
The combination of right-sizing, lifecycle cleanup, scheduled non-prod environments, fresh commitment coverage, and a newly negotiated EDP brought monthly spend down by roughly 25% — with no application changes, no re-architecture, and no disruption to production traffic.
Outcome
Before
$173,402
monthly AWS spend
After Koali
$130,189
monthly AWS spend
Annualized savings
Realized, ongoing
$518,556
ready when you are
What's in your bill?
Free audit. Read-only access. A dollar-backed findings report. The same process this case study started with.
$10k+ monthly AWS spend · No risk