The savings hidden in your Azure bill
Drop a Cost Management export (cost details or FOCUS, actual or amortized cost). We find the savings that billing data alone can already quantify: Reserved Instances and savings plans, Stop/Start, Azure Hybrid Benefit, disks, storage, Log Analytics. Resource by resource, in €/month.
Your file never leaves your computer
- The analysis runs entirely in your browser: nothing is uploaded, nothing is stored, no account needed.
- The page technically blocks every outgoing connection (
Content-Security-Policy: connect-src 'none'). You can even go offline once the page has loaded. - We never see your file or your results. This site sets no cookies and uses no analytics.
- In the Azure portal, open Cost Management, then Exports, at billing account or subscription scope.
- Create a Cost and usage details export (actual, amortized or FOCUS), in CSV or Parquet, for last month. Two or three months is even better.
- Run it (Run now), get the file from the storage account, then drop it here.
We recommend amortized cost: reservations are spread over the resources that use them. The file contains resource names and tags, which is exactly why it stays on your machine.
What your bill already tells
Monthly estimates, applied in cascade so the same saving is never counted twice. Amounts are orders of magnitude at list price: your negotiated discounts (EA, MACC) can shift them.
What your bill cannot tell
Billing data says what you pay, not how your resources are used. The levers below need usage metrics and the Azure inventory, and that is often where half of the savings are.
Rightsizing
A VM at 5% CPU costs as much as one at 90%. You need CPU and RAM at P95 over 30 days to go one size down safely.
Idle VMs
A VM that is running but idle (average CPU around 2%, almost no network out) looks exactly like a prod VM in the bill.
Unattached disks
The bill shows what a disk costs, not whether it is still attached to a VM. That state comes from the inventory (Azure Resource Graph).
Disk fill rate
A 1 TB P30 that is 10% full costs the same as a full one. Usage is read inside the VM, not in billing.
PaaS and databases
vCores, DTUs, App Service plans, Synapse, Fabric: without metrics, there is no way to tell whether they are oversized.
What is already committed
Existing reservations, MACC, licences with Software Assurance: to be checked before buying a new reservation.
An engagement goes further. With read-only (Reader) access to the metrics, we quantify these levers too, resource by resource, and tell you where to start.
Contact usHow the analysis works
Is my file sent anywhere?
No. The file is read by your browser (open-source libraries: PapaParse for CSV, hyparquet for Parquet) and aggregated in memory. No data is transmitted: the page security policy blocks every outgoing connection, which you can check in the Network tab of the developer tools. Closing the tab erases everything.
How are the savings estimated?
With deliberately cautious assumptions drawn from our engagements: 3-year Reserved Instance or savings plan ≈ 50% on an always-on VM; automatic Stop + Start on demand ≈ 90% on non-prod and VDI; Windows licence ≈ 40% of the OnDemand cost of a Windows VM (Azure Hybrid Benefit); Premium SSD to Standard SSD ≈ 50%; Hot to Cool tier ≈ 20% (half of the volume assumed cold); GRS to LRS on non-prod ≈ 50%; Log Analytics and Sentinel commitment tiers 15% to 35%; v1 / v2 VM to v5 ≈ 15%. Levers are applied in cascade on the same VM: Stop/Start, then Hybrid Benefit, then generation, then reservation.
Which exports are supported?
Cost details exports (formerly usage details) in EA or MCA schema, actual or amortized cost, as well as FOCUS exports (FinOps Open Cost and Usage Specification), versions 1.0 and 1.2-preview as produced by Cost Management. CSV files, gzip-compressed CSV (.csv.gz) or Parquet. The columns read include ResourceId, MeterCategory, MeterSubCategory, MeterName, ProductName, Quantity, UnitOfMeasure, CostInBillingCurrency (or Cost), PricingModel, ChargeType, Tags, BillingCurrency and Date. In FOCUS, their equivalents are read: EffectiveCost, ChargePeriodStart, ConsumedQuantity, PricingCategory, ChargeCategory, CommitmentDiscountStatus SkuMeter (or x_SkuMeterName in 1.0) and the Microsoft columns x_SkuMeterCategory, x_SkuMeterSubcategory, x_ResourceGroupName. Files re-saved by Excel (semicolon separator, decimal comma, dd/mm/yyyy dates) are recognised too.
How is the environment (prod, non-prod) determined?
First from the environment tag (env, environment…), then from the naming convention of the resource, resource group or subscription (dev, test, rec, uat, preprod, prd…). VDIs are identified by name or tags (avd, vdi, wvd).
What about AWS or GCP?
The tool reads Azure exports for now. For AWS (CUR) or GCP, write to us: we run the same analysis as part of an engagement.
How much is your cloud costing you for nothing?
A first 30-minute conversation to understand your context and see how we can help. No commitment.