Denode Systems. Simplifying the complex through intelligent systems.
Cloud overspend bills you every month. Our fee bills you once.
Two things happen when nobody owns cloud cost. Your console's recommendations pile up unactioned, and the architecture decisions behind the bill never get revisited. Over two weeks, for a flat fee, a person works through both, prices every finding, and hands you a report your team can act on the day it lands.
Illustrative. A constructed example, not a client bill.
| Compute Engine | $18,400 |
| BigQuery (on-demand) | $6,220 |
| Persistent Disk (unattached) | $4,215 |
| Cloud Storage (Standard) | $3,180 |
| Networking (egress) | $2,940 |
| Total | $34,955 |
Persistent Disk. Your console flagged this line in March and it is still billing, because working through the list has never won against shipping.
BigQuery. An hourly query scanning years of data when only yesterday changed. Nothing flags this one, because no rule asks whether the query should run that way at all.
The numbers, with their sources
of IaaS/PaaS cloud spend is wasted industry-wide, and 2026 is the first year in five that the figure went up.
Flexera, 2026 State of the Cloud Reporttypical waste where nobody owns cloud cost full time, against 10-15% with a staffed FinOps function.
Published 2026 FinOps maturity benchmarksfrom kickoff to a findings report in hand.
Denode Systems engagement standardIf nobody at your company owns cloud cost full time, the published spread puts you in the second group. That is who this audit is built for.
We do not publish a "typical % we recover" table. We have not audited enough estates to have our own benchmark, and someone else's average is not a prediction of your bill. It is exactly the kind of number we would tell you to distrust from a vendor, so your report will have your numbers in it instead.
Who this is for
Sound familiar?
Your GCP bill grew faster than your usage did.
Nobody has had a free week to find out why, and finance just sees the total climbing.
You suspect there's idle or forgotten infrastructure.
Cleanup never survives sprint planning against shipping features, so it sits there, quietly billing.
You want an independent second opinion.
Ahead of a renewal, a raise, or a board update, without hiring a full-time FinOps person for a one-time question.
Scope note: this is a generalist GCP cost review, built for SaaS and product companies.
The engagement, two weeks
How it works
Step 01
Kickoff & access
A short intake call to confirm scope, environments, and stakeholders. Then read-only access: billing first, plus a few viewer roles on resource metadata that make the findings sharper.
- We never touch production infrastructure and cannot change anything
- We read configuration and usage, never the contents of your storage objects or your tables
- The roles expire the day the engagement ends
Step 02
The data pull
We pull three to six months of billing exports and reconcile them against the invoices you paid.
- We log your committed-use, sustained-use and startup credits first
- So we never count savings you already have as a new finding
Step 03
The deep read
This is where most of the two weeks goes, and it has two halves. First the list your console has already built: the Recommender and the FinOps Hub have been flagging things for months, and we work through every one of them, evidence it and price it. Then the half nothing flags. A person reads the estate service by service and asks the question no recommender can ask: whether a resource should exist in that shape at all. Every finding in the report says which half it came from.
- BigQuery: what your queries really scan, on what schedule, and whether reserved capacity beats on-demand for your pattern. Usually the largest single finding
- GKE and Cloud Run: cluster and service shape, cross-zone traffic, warm capacity nobody is using
- Non-production that quietly became production: replicas, HA and always-on environments sized for traffic that never arrives
- Networking topology: inter-zone paths and egress by destination, where no per-resource rule has jurisdiction
- Logging and monitoring drift: ingestion volume, retention, and sinks writing the same logs twice
- Commitments priced against the baseline left after the cleanup, not the one you have today. Backwards is the most expensive mistake available here
Every finding points at a specific named resource, with the arithmetic shown.
Step 04
Findings & readout
You get a report ranked by payback, with every recommendation priced in dollars and effort. Then we walk your team through it live for an hour, and come back 30 days later to see whether the bill moved.
What you get
At the end of two weeks, you can answer the cloud-spend question in a board meeting.
Six deliverables, written for two readers at once. Enough detail for an engineer to start on Monday, and enough dollar clarity for finance to sign it off without a translation layer.
Ranked findings, priced individually
A monthly dollar figure and an honest effort estimate on every finding, sorted by payback. Each one is tagged with where it came from: already flagged by your console and not yet actioned, or not machine-detectable at all. Page one carries the split, so you can see what you paid for.
Before-and-after run rate
What you spend now, annualized, next to what it becomes if the recommendations land. One number for a board update.
A 30/60/90 action plan
Every item with an owner, an effort estimate, and its dollar impact, sized to drop straight into a sprint board.
Your commitment & credit position
Where your committed-use discounts, sustained-use pricing, and credits stand. What expires when, and what the bill does that day.
A 60-minute live readout
Your engineers and your finance lead in the same call, so the trade-offs get argued out in front of both. We record it, and the recording is yours.
A 30-day verification check-in
We go back to the real invoice a month later and confirm what moved. If something didn't land, you hear it from us, not from the next bill.
Everything is yours outright. There is no platform to stay subscribed to, and nothing stops working when the engagement ends.
Flat fee, no lock-in
Pricing
Starter audit
$2,500 flat
$5-10K/month in GCP spend
A single project or environment. The same four-phase process, scoped to one footprint.
- Full four-phase audit
- All six deliverables
- 60-minute live readout
Standard audit
$5,000 flat
$10-25K/month in GCP spend
Multiple projects or environments. The same depth, across the whole estate.
- Everything in Starter, across every project
- Commitment strategy for the next renewal cycle
Standard+
$7,500-10,000
$25K+/month, multi-BU or multi-org
We quote inside the range once we have seen the shape of the estate on the intro call.
- Everything in Standard
- Org-level policy and governance review
- Optional %-of-verified-savings follow-on
There is no retainer and no subscription. We agree the fee before we start and it does not move, whatever we find.
Not a dashboard, not a reseller
The people who sell you cloud are the wrong people to ask how to buy less of it.
A reseller earns margin on what you consume. That is a fine relationship for procurement and support, and the wrong one for a spend review. We resell nothing and take no vendor commission, so there is no version of this engagement where your bill going up is good for us.
Tooling has the opposite problem. A dashboard shows you a graph. It cannot read your architecture and explain why a disk has been idle since March, and it wants a monthly subscription either way.
The same person you meet on the intro call does the work. We hand nothing off to an account manager or a junior-analyst queue, and we ship the method alongside the findings, so you can check how we reached a conclusion instead of taking it on trust.
Before you book
Questions worth answering upfront
What access do you need, and can this break anything?
Read-only throughout. Billing Account Viewer is the one role we always need. Beyond that we ask for a few viewer roles on resource metadata and utilization, because without them the findings stay at SKU level and the architectural ones become unreachable. Nothing we ask for can change or break anything, none of it reads the contents of your objects or tables, and the roles expire when the engagement ends. Your team implements the recommendations on your own schedule, or we scope that separately if you want hands-on help.
Do we need to sign an NDA? How is our billing data handled?
Yes, happy to sign one before you grant any access. We use your billing data for this engagement and nothing else. We never share it or reuse it across clients, and we delete it once you have the findings and have confirmed them.
We already have committed-use discounts or startup credits. Is an audit still worth it?
Usually, yes. We log what you have already committed to and how much credit runway is left before anything else, and work around it. A lot of the value shows up in what to do when a commitment renews or credits run out, which is the point most internal reviews miss.
We already use a cost tool: CloudZero, Cloudability, Finout, or the native recommender. What does this add?
Those find what is findable by rule, and they are worth having. Two things this adds. The first is time: where nobody owns cloud cost, the recommendations pile up unactioned, because cleanup never wins a sprint against shipping. We work through that list, evidence each item and price it. The second is the question no rule asks, which is why an instance is oversized, or that a service is expensive because of an architectural decision made in 2024 that nobody has revisited. Every finding in the report says which of the two it came from.
Are you a Google partner or reseller? Will this affect our existing partner relationship?
No to the first, deliberately. Partners and resellers earn margin on what you consume, which makes them the wrong people to ask how to consume less. We take no vendor commission and have no reseller agreement with Google, so a flat fee is the only money in this for us. And no to the second: this is buyer-side advisory that sits alongside whatever you already have. Your partner keeps the contract, the support relationship, and any negotiated pricing.
Do you help implement the changes, or just hand over a report?
The core engagement is the findings, the roadmap, and the readout. That is what keeps the incentive on finding real waste rather than billing more hours. Implementation support is available, scoped and quoted separately, if you would rather not have your own team do it.
What if you don't find meaningful savings?
We report thin findings honestly. The flat fee buys the read, not a guaranteed number. Industry-wide, 29% of cloud spend is wasted, and the figure runs higher where nobody owns cost full time, so a clean estate is uncommon. It happens, though, and "your setup is already tight, here is the evidence, don't spend money on this again next quarter" is a real and useful answer. You can take that to a board with more confidence than a number we had bent to justify the invoice.
How fast can we start?
Usually within a week of the intro call, subject to current capacity.
Get a second opinion on your GCP bill.
Thirty minutes, no obligation. Bring your last invoice if you have it handy.
Prefer email?
Reply in 1 business day.
We reply from a real inbox, not an autoresponder. By sending this you agree to our privacy notice. Plain version: we use it to reply to you, and nothing else.