Spend you can explain to the board.
Cost work fails when it is a one-off spreadsheet exercise. We install the reporting, the tag policy and the review rhythm, then keep pulling the levers month after month.
We charge a fixed fee, never a share of savings
A percentage-of-savings model pays a provider more for finding waste slowly, and rewards leaving some behind for next quarter. It also makes the provider's interests diverge from yours the moment the easy wins are gone.
Our FinOps work sits inside the fixed monthly service fee. We also never mark up your AWS consumption — if you bill through us, you pay our cost.
What a first quarter typically returns
Ranges observed across our customer base. Actual results depend on workload profile and existing commitments — these overlap, so they do not simply add up.
Pulled in this order, for a reason
Buying commitments before you rightsize locks in the wrong baseline for three years. Order matters more than effort.
Delete what nobody owns
Orphaned volumes, unattached addresses, idle load balancers, forgotten non-production environments. Zero architectural risk, immediate return.
Rightsize against real usage
Not the p50, and not the peak from eighteen months ago. Graviton and modern instance families where the workload supports it.
Fix the storage tiering
S3 lifecycle policies, Intelligent Tiering where access is unpredictable, EBS gp2 to gp3, and snapshot retention that matches the actual policy.
Then buy commitments
Savings Plans and Reserved Instances sized against the post-rightsizing baseline, laddered so you are not renewing everything in the same month.
Redesign the expensive paths
Cross-AZ chatter, NAT gateway egress and inter-region transfer are quietly among the largest line items on most estates.
Attack observability spend
Log retention, custom metric cardinality and third-party ingest pricing. Often 5–10% of a bill, and rarely reviewed by anyone.
Attribution first, or none of it sticks
You cannot ask a team to reduce a number they cannot see. Before optimisation we make spend attributable — which is usually the harder half of the work and the reason it holds after we finish.
- Tag policy enforced with SCPs and Config rules, not a wiki page
- Account structure aligned to product or business unit, so the boundary does the attribution for you
- Showback dashboards that go to engineering leads, not just finance
- Unit economics where it is meaningful — cost per tenant, per transaction, per active user
- Anomaly detection routed to the team that caused the change, within hours rather than at month end
- Budget guardrails in CI, so a Terraform plan shows the cost delta before merge
Untagged spend is the tell
On first engagement we typically find 30–45% of spend untagged. Getting that under 5% is worth more over two years than any single rightsizing exercise, because it is what makes the next round possible.
Optimisation is a habit, not a project
Spend review
Trend by account, team and service; variance against forecast; what changed and who changed it; the next three actions with owners.
Commitment review
Coverage and utilisation against target, upcoming expiries, and a laddered purchase recommendation sized to the current baseline.
Anomaly response
Detection tuned to your patterns so it catches the real thing without crying wolf, routed to the owning team the same day.
“They found $190k of annual waste in the first fortnight and then refused to bill us a percentage of it. That told me everything.”
The negotiation, not just the engineering
A large share of cloud savings is commercial rather than technical. We support the conversations most engineering teams are not equipped to have, and most procurement teams lack the usage data for.
- Enterprise Discount Program modelling and negotiation input
- Marketplace spend that counts toward commitment thresholds
- Invoice reconciliation and credit tracking
- Migration funding applied for and tracked to receipt
- Multi-year commitment risk modelling against your growth plan
What we need to start
- Read-only access to Cost Explorer and the Cost and Usage Report
- Read-only access to the accounts in scope
- Ninety minutes with whoever owns the largest three workloads
That is enough for a first findings session inside two weeks. We do not need production write access to tell you where the money is going.
Request a cost reviewFinOps questions
Will optimisation make our platform less reliable?
Not if it is done in the right order. Every recommendation carries a risk rating, and anything touching redundancy or headroom goes through the same change process as any other production change. We will not trade an availability commitment for a cost number without you deciding that explicitly.
We already use a cost tool. Do we need you?
Tools show you the number. They do not restructure your accounts, negotiate your commitments, or make an engineering team act on a dashboard. We are happy to work inside a tool you already own rather than sell you another one.
How quickly do savings appear?
First actions typically land within two weeks and show on the next invoice. Commitment changes take effect immediately but reveal their full value over the term. Structural work — account restructure, architecture change — lands over one to two quarters.
What if our spend is genuinely efficient already?
Then we tell you, in writing, and we do not manufacture findings to justify a fee. It happens. In those cases the value is usually in attribution and forecast accuracy rather than reduction.
Find out what is actually in your bill
A first findings session in two weeks, from read-only access. No production changes required to get the answer.