Customer story · E-commerce · Anonymized
45% cost reduction
on monthly AWS spend
Refine surfaced storage-class moves we had missed for two quarters. The Optimization Score paid for our forecast accuracy alone.
By HabileLabs
$30k
Monthly AWS spend (before)
$24.75k
Monthly AWS spend (after)
$243k
Annualized savings
14 days
Time to first impact
E-commerce · This story is anonymized at the customer's request. Numbers and outcomes are real; identifying details have been changed.
The problem
What they were dealing with
A fast-growing e-commerce platform was scaling quarter over quarter. Monthly AWS spend had climbed from $18k to $45k over 14 months without a proportional revenue lift. The finance team had stopped asking why — they assumed scale was the answer. Engineering knew better but had no time to investigate.
A consultant flagged that S3 storage costs looked unusually high relative to traffic. The same week, three EC2 instances appeared on the bill that nobody on the on-call rotation recognized. The team needed a single view of cost and security without spinning up a dedicated FinOps function.
The approach
How they used Refine
The team connected Refine via the read-only IAM role. The first narrative report ran in under a minute. Three observations surfaced in priority order: untiered S3 storage costing $2.4k/mo more than it should, two over-provisioned EC2 instances running 24/7 for batch jobs that finish in 4 hours, and a missing lifecycle policy on a log archive bucket.
Over the next two weeks, the engineering team worked through the top 6 recommendations. Each came with the AWS API change in plain English and the projected savings range. Where remediation was safe and obvious — enabling EBS encryption defaults, attaching a missing S3 lifecycle policy — the "Fix Available" hook applied the change in a click.
- Connected Refine via read-only IAM role in 60 seconds
- Surfaced top 6 recommendations ranked by monthly dollar impact
- Implemented S3 lifecycle to move cold logs to Glacier ($2.4k/mo)
- Right-sized 2 over-provisioned EC2 instances ($3.1k/mo)
- Set up auto-scaling for variable workloads ($1.8k/mo)
- Applied EBS encryption defaults org-wide via Fix Available (opt-in scoped policy)
The result
45% cost reduction · $45,000 → $24,750 / month
After 14 days, the monthly AWS bill had dropped from $45k to $24.75k — a 45% reduction. The Optimization Score moved from 41/100 to 76/100. Just as important, the team unmuted a category of security findings that had been deferred while cost was on fire, and closed three Critical-severity IAM issues that would have surfaced later in an audit.
Six months later, the team uses Refine as its weekly cost ritual. The narrative report goes to engineering leads every Monday morning. Anomaly alerts have caught two runaway Lambdas before they made it into a billing cycle. Allocation tag coverage climbed from 41% to 89%, finally making per-team showback workable.
"Before Refine, half our prod EC2 went to "Unallocated" on the showback report. After two weeks of tag cleanup with the untagged-by-spend list, we had real attribution — and the savings work could finally target the right teams."
— Engineering lead, E-commerce (anonymized)
See what Refine surfaces in your account
60-second setup. Free forever. Read-only access.
Refine is built and supported by HabileLabs, an AWS Advanced Tier Services Partner.