GreenOps metrics are the numbers a team uses to track the energy, carbon and water behind its cloud and AI usage, reported next to cost. The usual set is the total cloud carbon footprint in kg or tonnes CO2e, energy in kWh, water in litres, carbon per unit of business output, carbon waste and the share of carbon emissions allocated to a team or product. A metric becomes a KPI once it has an owner, a target and a review date.
Key takeaways
- Track energy, carbon and water at the same level of detail as the cloud bill.
- Make carbon per unit of business output the main KPI, because total emissions rise with growth.
- Treat carbon per unit of spend with care, because discounts change cloud costs without changing energy use.
- Start with a baseline and two KPIs, and add targets once the data is trusted.
What are GreenOps metrics?
GreenOps applies the habits of FinOps to environmental impact, so where FinOps teams track cloud cost by account, service and team, GreenOps, DevOps and engineering teams track the energy, carbon and water behind the same usage.
Most GreenOps metrics start from the billing exports FinOps teams already use, such as the AWS Cost and Usage Report, which lists every line of usage with a quantity, a region and a timestamp. A data provider such as Greenpixie estimates the energy for each line, converts it to CO2e using the carbon intensity of the local grid, and converts it to water using regional water factors.
GreenOps metrics vs GreenOps KPIs
A metric is a measurement, such as total cloud emissions last quarter, and a KPI is a metric the business has chosen to manage, with a named owner, a target and a date to review it. One Greenpixie customer, a global technology services company, turned its cloud carbon metric into a KPI when the CIO gave the FinOps team a 10% cloud carbon reduction target on top of its cost goals.
The FinOps Foundation's sustainability capability describes success in similar terms, with carbon reported next to cost, split by team and product, and tracked against carbon budgets.
Common GreenOps and sustainability metrics
The table lists the metrics most GreenOps programmes start with, in the units the FinOps Foundation's sustainability capability uses.
| Metric | Unit | What it shows |
|---|---|---|
| Cloud carbon footprint | kg or tonnes CO2e | Total emissions from cloud usage over a period, and the baseline for every other figure |
| Energy consumption | kWh or MWh | Electricity used by the infrastructure behind the usage. Energy tracks cost more closely than carbon does |
| Water consumption | Litres or m³ | Water used to cool the data centres behind the usage |
| Carbon per unit of usage | kg CO2e per vCPU-hour, GB-month or other pricing unit | Efficiency of a resource type, for comparing cloud services, instance families, storage tiers and regions |
| Carbon per unit of business output | g CO2e per transaction, active user, API call or 1,000 tokens | Whether efficiency improves as the business grows |
| Carbon per unit of spend | kg CO2e per £1,000 or $1,000 of cloud spend | A link between cost and carbon reports. Discounts distort it |
| Carbon waste | kg CO2e | Emissions from resources sized beyond what a workload needs |
Cloud carbon footprint, energy and water
Carbon and water are both calculated from energy, with carbon depending on the grid in each region at the time of use and water on how each data centre is cooled, so a region move that cuts carbon can put a workload somewhere water is scarce. Some methods also add embodied carbon, the emissions from manufacturing the servers, shared across their working life. Greenpixie's cloud methodology sets out how each layer is calculated.
Carbon per unit metrics
Unit metrics, the sustainability side of unit economics, separate growth from efficiency by dividing carbon emissions by something the business already counts, such as orders, active users or AI tokens. If transactions double while emissions rise by 40%, carbon per transaction has fallen by 30%.
The Green Software Foundation's Software Carbon Intensity (SCI) specification gives software development teams a standard unit metric. It adds a system's operational emissions, energy multiplied by carbon intensity, to its share of embodied emissions and divides the total by a functional unit. It is published as ISO/IEC 21031:2024.
Carbon per unit of spend is quicker to produce, but a savings plan or reserved instance cuts the bill without changing the infrastructure or energy behind it, so the ratio rises when nothing physical has changed. We suggest setting efficiency targets on usage or business output instead.
Carbon waste, efficiency and optimisation
The FinOps Foundation defines carbon waste as the emissions of the resource currently allocated minus those of one sized for actual utilisation, and carbon efficiency as the inverse. Both come from the same analysis as rightsizing, so each rightsizing recommendation can carry a CO2e figure. Organisations estimate that 29% of their IaaS and PaaS spend is wasted, according to Flexera's 2026 State of the Cloud Report, and idle or oversized cloud resources still draw power, so cloud waste carries emissions as well as cost.
Data centre metrics for cloud customers
Data centre metrics describe cloud infrastructure outside a customer's control, but they feed into the emissions and water figures for its usage.
- Power usage effectiveness, or PUE, is total facility energy divided by IT equipment energy. The industry average was 1.52 in 2026 according to the Uptime Institute's 2026 survey, and AWS, Microsoft and Google publish fleet figures between 1.09 and 1.17. Our guide on how to calculate PUE has the formula and provider figures.
- Water usage effectiveness, or WUE, is litres of water per kWh of IT energy. It changes with cooling design and climate.
- Grid carbon intensity is grams of CO2e per kWh of electricity in a region. It changes by region and by hour, which carbon-aware computing uses to move flexible workloads to cleaner times and places, and the same AWS m5.12xlarge instance running for a year produces about 620 kg CO2e in Hong Kong, region ap-east-1, and 284 kg in Spain, region eu-south-2, which is 54% less, according to Greenpixie data from November 2026.
FinOps carbon metrics and KPIs
FinOps and GreenOps work from the same billing data, and FinOps teams usually own GreenOps and carbon reporting, because they already own the billing data, the allocation model and the reviews with engineering. These KPIs follow the FinOps Foundation's measures of success, and each can be calculated from billing data enriched with carbon.
| KPI | How to calculate it | Usually owned by |
|---|---|---|
| Carbon per unit of business output | kg CO2e divided by transactions, users or 1,000 tokens, tracked monthly | Product or platform team |
| Allocation coverage | Emissions allocated to a team or product divided by total cloud emissions | FinOps team |
| Recommendations with a carbon figure | Optimisation recommendations that show cost and carbon impact divided by all recommendations | FinOps team |
| Carbon removed through optimisation | Carbon waste removed by completed rightsizing, clean-up and scheduling work | Engineering and DevOps teams, reported by FinOps |
| Emissions forecast variance | Actual emissions minus forecast emissions, divided by forecast emissions | FinOps and finance |
| Progress against carbon budget | Emissions to date divided by the carbon budget for the period, per team or product | Team owners, with the sustainability team |
We suggest starting with carbon per unit of output and allocation coverage, then adding the others once each KPI has an owner who reviews it every month, next to cost and performance.
KPIs for each stage of the GreenOps maturity model
The FinOps Foundation's maturity model has three stages, Crawl, Walk and Run, and its sustainability capability describes each one. Our guide to FinOps sustainability covers what the stages look like in practice. We suggest these KPIs as GreenOps practices mature:
- Crawl. Impact assessment happens after deployment, through periodic reports. Track a 12-month baseline by provider, account and region, and allocation coverage.
- Walk. Emissions appear in allocation and showback, and in changes to architecture. Add carbon per unit of output, the share of recommendations with a carbon figure and carbon removed through optimisation.
- Run. Emissions are part of every migration, development and optimisation decision across the organisation. Add carbon budgets per team, forecast variance and progress against company targets.
Why carbon KPIs need usage-based data
Spend-based methods multiply cloud spending by an emission factor per pound or dollar, so a discount moves the emissions figure even when energy use stays the same, and a KPI built on them rewards negotiating a better rate. Usage-based methods estimate energy for each line of usage and convert it to carbon.
Provider carbon dashboards are free, but each uses its own method and level of detail, and their figures are hard to split by tag or product. Our guide to Cloud Carbon Footprint alternatives sets out the options side by side.
Each KPI needs a stated emissions basis that stays the same each year. Location-based figures follow the grid in each region, while market-based figures reflect a provider's renewable energy contracts, and the GHG Protocol Scope 2 Guidance asks companies to report both. Cloud emissions belong in a customer's Scope 3, which our guide to Scope 3 cloud emissions covers.
Setting up GreenOps KPIs
- Get carbon data at the level of the bill. Load energy, carbon and water for each line of cloud billing data into the FinOps or cost management tool the team already uses. GPX Cloud does this from CUR or FOCUS exports with read-only access, and Greenpixie's methodology has been independently verified in accordance with ISO 14064-3, prepared with reference to ISO 14064-1, and designed to be in line with the GHG Protocol Corporate Standard.
- Set a 12-month baseline. Use the same boundary and method the company uses for its GHG reporting, and write the method down. The baseline is also the starting point for any business case.
- Choose two KPIs to start. Use carbon per unit of business output and allocation coverage.
- Give each KPI an owner and a review date. Put emissions into the monthly cost reviews, optimisation recommendations and cost alerts the team already runs.
- Set targets once the data is trusted. Split the organisation's sustainability goals into carbon budgets per team or product, and add emissions to forecasts.
"Engineers typically don't really care that much about cost when their KPIs [are] around performance. So when you're putting that sustainability data in front of them, you are kind of changing the way they view their usage and making the units essentially more meaningful to them."
GreenOps metrics in practice
- Mastercard. Cloud cost optimisations had stalled until carbon data on the same recommendations gave engineers a second reason to act. Greenpixie data enabled $2M in cloud cost reductions, and Mastercard exceeded its cloud carbon reduction target.
- AI usage. In a case study from Greenpixie's Cloud GreenOps Philosopher course, a team baselined its token usage and then changed how it used models, cutting CO2e per token by 82% and cost per token by 59%. Our guide to sustainable AI covers per-token measurement.
- Storage. In another case study from the course, moving data to the right storage tiers cut storage cost by around 19% and storage emissions by around 60%, a gain a cost-only KPI would have missed.
To see carbon next to cost for your own cloud bill, get started for free in the Greenpixie app. It adds energy, carbon and water to AWS, Azure and Google Cloud billing data.
Frequently asked questions
What is an environmental KPI?
An environmental KPI is an environmental measure, such as tonnes of CO2e or litres of water, that a business manages against a target with a named owner. For cloud teams, a common one is carbon per unit of business output.
What are examples of GreenOps KPIs?
Carbon per transaction or per 1,000 AI tokens, the share of cloud emissions allocated to teams, the share of optimisation recommendations that show CO2e, and progress against a team carbon budget. A CIO-set target, such as a 10% cut in cloud carbon, is another.
What does GreenOps stand for and what does it mean?
GreenOps stands for green operations. It applies FinOps practices to the energy, carbon and water behind cloud operations and AI usage, so teams manage environmental impact with the same data and routines they use for cost. Our guide to what GreenOps is covers the meaning, principles and examples.
Is there a framework for GreenOps KPIs?
The FinOps Foundation's sustainability capability is the closest to a standard for cloud sustainability KPIs, with measures such as carbon per unit, carbon waste and forecast accuracy, and a description of each maturity stage. For per-unit software metrics, the SCI specification, ISO/IEC 21031:2024, gives a standard formula.
How do I implement GreenOps?
Get carbon data at the same level of detail as your cloud bill, set a baseline and add emissions to the cost reviews your FinOps team already runs. Then pick two KPIs, give each an owner, and set targets once the data is trusted.
How do I set a baseline for cloud emissions?
Total energy, carbon and water for the last 12 months of usage across all cloud environments by provider, account, region and team, using the same boundary and emissions basis as your company's GHG reporting. Write the method down so later figures compare like for like. Our guide to sustainable cloud computing covers the wider picture.
Explore GreenOps
Put carbon next to cost
See carbon next to cost for your own cloud bill
GPX Cloud adds energy, carbon and water to AWS, Azure and Google Cloud billing data.
Sources
- FinOps Foundation, Sustainability capability in the FinOps Framework
- FinOps Foundation, FinOps maturity model
- Green Software Foundation, Software Carbon Intensity (SCI) specification
- IEC, ISO/IEC 21031:2024 Software Carbon Intensity specification
- Flexera, 2026 State of the Cloud Report
- Uptime Institute, Global Data Center Survey 2026
- GHG Protocol, Scope 2 Guidance
- Greenpixie Cloud methodology
- GreenOps Academy, Cloud GreenOps Philosopher course
- Capgemini, Cloud Realities podcast transcript with James Hall

