Cloud Cost Optimization
Roughly a third of most cloud bills is idle or oversized. We start there, not with your architecture.
Book a cost review ↗
Why is our cloud bill so high?
Cloud cost optimization is mostly housekeeping, and the first round of it is dull. In nearly every estate we review, something close to a third of the bill is resources nobody is using: staging environments running through the weekend, instances sized during a launch two years ago, orphaned volumes and snapshots, load balancers pointing at nothing.
The interesting work — caching layers, spot capacity, re-architecting a chatty service — only pays once the obvious waste is gone, and it spends engineering time you may want elsewhere. So we go in that order: delete, then right-size, then commit, and only then talk about design. Most clients are happy to stop after step three, and we’d rather not re-platform anything if we don’t have to.
What a cost review covers
Full spend breakdown
Twelve months of billing data cut by service, environment and team, so the conversation starts with where the money actually goes.
Idle resource sweep
Unattached volumes, stale snapshots, idle databases, and dev environments billed for 168 hours a week to serve about 40 of use.
Right-sizing pass
Measured CPU, memory and IOPS against what’s provisioned, with a recommended size per workload and the risk of each change stated.
Commitment modelling
Reserved instances and savings plans modelled against your true baseline only — we don’t commit you to capacity you might not want in a year.
Storage and egress
Lifecycle rules, tier moves, and the data-transfer paths that quietly cost more than the compute they exist to serve.
Guardrails that hold
Tagging, per-environment budgets and alerts, so the savings don’t erode the next time someone spins up a test cluster and forgets it.
How a cost review runs
Read-only access first
Billing exports and read-only roles on the accounts. Nothing gets changed or switched off in this phase, and nothing at all without your sign-off.
Find the idle money
The unglamorous sweep: orphaned resources, non-production schedules, duplicate environments. This is usually where the biggest single number comes from.
Right-size what stays
We resize against measured load in small batches, watching latency and error rates after each one, and reverse anything that degrades.
Commit the baseline
Once the estate has settled at its real shape, we model savings plans against it — committing early is how teams lock in yesterday’s waste.
Leave the guardrails
Tagging, budgets and a monthly one-page review your team can run, plus a written list of anything we deliberately left alone and why.
Cloud Cost Optimization FAQ
A review of a single estate is typically $5k–$15k depending on account sprawl and how many services are involved. If the first week doesn’t surface savings worth several times the fee, we’ll say so and stop rather than pad the report.
The tools we build with
We measure before we touch anything — the cheapest saving is always the resource you can prove nobody is using.
Cloud & billing
AWS
GCP
CloudflareFinOpsMeasurement
Prometheus
Grafana
OpenTelemetry
DatadogChange control
Terraform
Ansible
Kubernetes
DockerAnalysis
Python
PostgreSQL
Snowflake
TypeScriptRelated work
Related reading
Bill grown faster than your traffic?
Give us read-only access and last year’s invoices. We’ll come back with a ranked list of savings, what each one is worth, and what it risks.