How we reduced AWS infrastructure costs by 18% without replatforming

We audited a growing AWS infrastructure service by service, removed unused resources, and cut the monthly bill from $17,331.69 to $14,266.71. No replatforming, no disruption to operations.

A donation platform for charitable organizations runs on AWS. Over time, the infrastructure grew faster than the routines for keeping it lean, and the monthly bill grew with it. In three months, we brought it down 18% without replatforming or disrupting the operation. This is how.

linkThe challenge

The client operates a donation platform for charitable organizations, built on AWS. As the product evolved, infrastructure was added to meet each new need, but there was no continuous optimization process running alongside that growth.

The result was a pattern we see often: costs rising month over month, with nobody able to point to exactly why. The problem was not the cloud provider, and it did not call for a migration or a redesign. It was a control problem.

linkWhat we found

A service-by-service audit surfaced four sources of silent spend:

1. Unused services still running. Resources provisioned for past needs that nobody had decommissioned, still generating charges every month.
2. Forgotten regions. Infrastructure deployed in AWS regions that were no longer part of the operation, quietly billing in the background.
3. Oversized plans. Resources running on more expensive tiers than their actual usage required.
4. Obsolete components. Accumulated pieces that no longer served the product but complicated maintenance and added technical debt.

Starting monthly cost: $17,331.69.

linkWhat we did

The work was deliberately conservative. Nothing was deleted on a hunch, and nothing was removed without a way back:

• A complete audit of the AWS infrastructure, service by service, to map every source of cost against actual usage.
• Controlled removal of unused resources, in stages rather than in bulk.
• Migration to more economical plans wherever real usage justified it.
• Identification and shutdown of forgotten regions that were still generating costs.
• A safe rollback strategy using Glacier: before any resource was permanently deleted, it was archived, so any removal could be reversed if the operation needed it back.

That last point matters more than it looks. Cost cleanup done carelessly creates outages. Cost cleanup done with a rollback path creates savings and keeps them.

linkResults

After three months, the infrastructure was measurably cleaner, more modern, and easier to maintain. Monthly costs went from $17,331.69 to $14,266.71:

• More than $3,000 USD saved every month, equivalent to over $36,000 annually if maintained at the new baseline
• A simplified infrastructure with fewer components to maintain
• Lower technical debt
• Better visibility and control over future costs

linkThe takeaway

Infrastructure optimization does not always require changing providers or redesigning the architecture. In many cases, the biggest savings come from something less dramatic: a periodic, disciplined review of the resources you already have. The spend that hurts most is usually the spend nobody is looking at.