Key takeaway
- Multi-cloud FinOps with FOCUS: what the spec standardises, which columns to use first, allocation strategy and a six-step adoption sequence.
Multi-cloud FinOps works when every provider's bill can be read in the same shape. The FOCUS (FinOps Open Cost and Usage Specification) is an open specification for generating consistent technology billing datasets, so a CFO, a platform lead and a FinOps practitioner can compare AWS, Azure, Google Cloud and other vendors without rebuilding each schema by hand. The FinOps Foundation defines FinOps as an operational framework and cultural practice that maximises the business value of technology, enables timely data-driven decision making and creates financial accountability through collaboration between engineering, finance and business teams (source: finops.org).
This guide takes the multi-cloud allocation angle: what FOCUS standardises, which columns to use first, how it supports showback and chargeback, and where it does not help. For commitment discounts, AI workloads and data platforms we keep separate guides, such as Savings Plans for cloud FinOps and FinOps for AI adoption costs.
What is FOCUS and why does it matter for multi-cloud FinOps?
According to the FOCUS site, it is an open specification that normalises billing data across AI, cloud, SaaS, data centre and other technology vendors, supported by the FinOps Foundation. It sets requirements for billing data generators, so that each provider's export uses the same column names and meanings. Without it, a team that runs three clouds typically maintains three schemas and three sets of transformation logic before any analysis starts. The Foundation's Data Ingestion capability states that FOCUS "delivers consistency and standardization to cloud cost data", which otherwise uses proprietary provider schemas.
The specification has moved quickly. The table below uses the ratification dates published on the FOCUS site.
| Version | Ratified | What it adds (per the FOCUS site) |
|---|---|---|
| 1.1 | 7 November 2024 | Columns for cloud service provider data, better metadata for ETL processes and some normative changes |
| 1.2 | 29 May 2025 | SaaS, PaaS and cloud billing in one schema, links from charges to invoice IDs, multi-currency and unit-cost support |
| 1.3 | 5 December 2025 | A Contract Commitment dataset, allocation columns, data recency and completeness flags, Service Provider and Host Provider columns |
| 1.4 (latest) | 4 June 2026 | 2 datasets, 47 columns, 6 attributes, 17 glossary entries and 2 supported features, with no incompatible changes |
Provider support is not uniform. The FOCUS home page lists AWS, Microsoft (Azure) and Google Cloud datasets at version 1.2 and Oracle Cloud at 1.3, with some other vendors at 1.0 or 1.3. The page does not say whether these are each provider's latest versions, so check the version in the files you actually ingest.
Which FOCUS columns should a multi-cloud team use first?
The specification defines the data that must appear in a FOCUS file and how columns are named. The Column Library is the full reference; these are the ones that matter first for allocation and finance reconciliation.
| Column group | Columns | Why it matters across clouds |
|---|---|---|
| Cost | Billed Cost, Effective Cost, List Cost, Contracted Cost | Billed Cost is the cost as invoiced in a billing period; Effective Cost is the cost of a charge based on resources and services consumed plus any contract commitments booked within the charge period. Together they let finance reconcile invoices while engineering sees the amortised cost of usage |
| Allocation | Allocated Method ID, Allocated Method Details, Allocated Resource ID, Allocated Resource Name, Allocated Tags | Record how a shared charge was split and to whom, instead of hiding the split in a spreadsheet |
| Provider | Service Provider Name, Host Provider Name | Separate who sold the service from who hosts it, which matters once SaaS and marketplace charges sit beside cloud bills |
| Tags | Tags | Carry the ownership metadata that allocation depends on, in one column rather than provider-specific label formats |
Billed Cost and Effective Cost can differ when commitment purchases are recorded separately from the charges they cover. That is why a commitment-heavy estate should report both; see the capacity reservation model for the commitment side. Column definitions are from the FOCUS 1.4 specification.
How do you allocate multi-cloud spend with FOCUS?
FOCUS standardises the data; the Foundation's Allocation capability describes the decisions you still have to make. It names three strategies that apply across providers.
| Strategy | What you decide | Multi-cloud consideration |
|---|---|---|
| Allocation strategy | How costs map to the organisation for showback, chargeback or cost-centre reporting | Finance, engineering and operations often need different views of the same spend, so several layers can coexist |
| Tagging and metadata strategy | How resources are split into accounts, how accounts are grouped, naming standards and how tags or labels are applied and aggregated | Must cover compliance gaps, differences between environments and costs that cannot be tagged |
| Shared cost strategy | How support, central networking or shared platforms are funded | Options include central budgeting or allocation by fixed, proportional or proxy-based methods; most organisations mix them |
The Foundation's maturity direction runs from recording ownership manually after deployment (Crawl), to putting ownership fields in provisioning templates (Walk), to enforcing metadata as a precondition for provisioning (Run). It calls this "Shifting Left". For a multi-cloud estate, that means one metadata standard in your infrastructure-as-code modules for every provider, not one tagging policy per cloud. Our view: this is where most multi-cloud allocation effort is won or lost, because FOCUS can only carry tags that were applied. See also FinOps on Google Cloud and FinOps for data management for provider and platform specifics.
How do you adopt FOCUS in your organisation?
Adoption is mostly a data engineering and operating-model exercise. A sequence that fits the Foundation's Data Ingestion guidance:
Inventory your sources: list every provider, SaaS vendor and platform whose bill you need, and note which already export FOCUS-format data and at which version.
Ingest and normalise: load FOCUS exports into one repository; where a provider has no native export, use a converter. The Data Ingestion capability lists FOCUS converter exports for AWS, Azure, Google Cloud and Oracle as inputs.
Check completeness: FOCUS 1.3 added data recency and completeness flags; use them so finance does not close a month on data that is still changing.
Fix allocation at source: apply one ownership and metadata standard in provisioning templates and enforce it in the pipeline.
Standardise reporting: agree definitions for Billed Cost and Effective Cost, who uses which, and the showback or chargeback calendar.
Measure the pipeline: the Foundation suggests KPIs such as Cost Visibility Delay, Frequency of Data Updates and ETL Processing Time, plus checks that provider data arrives on schedule and that FOCUS validator checks pass.
FOCUS Framework
The FOCUS framework emphasizes clarity and alignment in financial operations across multiple cloud environments.
Performance Measurement
Establishing KPIs and metrics to assess the effectiveness of FinOps strategies aids in continuous improvement.
Video: XenonStack's FinOps solution for AWS, one provider view of the cost visibility work described above.
What are the trade-offs, and when is FOCUS not enough?
| Situation | What FOCUS gives you | What it does not solve |
|---|---|---|
| Mixed provider versions | A common column model across AWS, Azure, Google Cloud, Oracle and others | Vendors publish at different versions (1.0 to 1.3 on the FOCUS home page), so columns and features can differ between files |
| Poor tagging | A single Tags column and allocation columns | Missing or inconsistent tags upstream; FOCUS cannot invent ownership |
| Commitment discounts | Billed Cost and Effective Cost, plus a Contract Commitment dataset from 1.3 | The purchasing decision itself; see Savings Plans for that |
| Carbon and sustainability | Billing data only | Carbon metrics need other data; see GreenOps KPIs. The Foundation expects FOCUS to extend to sustainability data, but that is not part of the billing columns above |
| AI and token spend | FOCUS 1.2 added support for virtual currency burn-down | Workload-level engineering decisions; see FinOps for AI engineering |
Our view for agentic operations: when software agents begin to propose rightsizing, tag fixes or budget alerts, consistent FOCUS data gives them one input shape across providers. The actions themselves still need policy limits and a decision trace. ElixirData Context OS keeps governed context, versioned policy and decision traces in one decision layer above AI agents, and AgentAnalyst lets teams ask questions of enterprise data in plain language and get real-time insights.
Who owns what in multi-cloud FinOps?
| Role | Responsibility (per the Foundation's Allocation capability) |
|---|---|
| FinOps practitioners | Develop naming and compliance standards, coordinate with other teams and evaluate allocation compliance |
| Finance | Decide the organisational and budget units for allocation and how shared costs are split |
| Engineering | Apply and automate metadata standards and supply metadata sources |
| Leadership | Approve allocation strategies and set the required granularity |
For a wider view of the discipline, the cloud-native FinOps services page and our guide to DataFinOps cover the operating model.
To map one cost decision your agents should make, and the context and policy it depends on, book a Context Workshop. See what happens in the workshop.
Frequently asked questions
What does FOCUS stand for?
FOCUS stands for FinOps Open Cost and Usage Specification. It is an open specification that normalises billing data across cloud, SaaS, AI, data centre and other technology vendors, and it is supported by the FinOps Foundation.
What is the latest version of FOCUS?
According to the FOCUS site, version 1.4 is the latest, ratified on 4 June 2026. Version 1.3 was ratified on 5 December 2025 and 1.2 on 29 May 2025.
What is the difference between Billed Cost and Effective Cost?
Billed Cost is the cost of a charge as invoiced by the invoice issuer in a billing period. Effective Cost is based on the resources and services consumed plus any contract commitments booked within the charge period. They differ when commitment purchases are recorded separately from the charges they cover.
Does FOCUS replace tagging?
No. FOCUS carries tags and allocation details in standard columns, but ownership metadata still has to be applied to resources. The FinOps Foundation recommends enforcing metadata at resource creation.
Do all cloud providers export FOCUS data?
Many do, but at different versions. The FOCUS home page lists AWS, Microsoft (Azure) and Google Cloud at 1.2 and Oracle Cloud at 1.3, with others at 1.0 or 1.3, so check the version of each file you ingest.
Follow XenonStack on LinkedIn: XenonStack on LinkedIn for new guides on agentic AI, data and platform engineering.
Ready to put AI agents to work? Explore ElixirClaw, the agentic OS platform to build, orchestrate and govern AI agents.
Explore the XenonStack platform family: ElixirData · ElixirClaw · ElixirHub · NexaStack · MetaSecure · Neural AI · AgentGRC · AgentSRE · AgentAnalyst · AgentInstruct · AgentRAI · AgentScrum · AgentSearch · AgentForce · AI Aviator · SAIF Aviator · CohortIQ · XenonStack AI · XenonStack UAE