7 Key ESG Metrics Every Company Should Track (2026 Checklist)

ESG metrics are the indicators a company uses to measure environmental, social and governance performance, from Scope 1 to 3 emissions and energy use to the gender pay gap and board independence. Most ESG dashboards carry 40 or more of them and prove nothing. Seven carry the regulatory and investor weight; the rest is documentation.

This guide gives the definition, a full list of the common metrics mapped to the standards that ask for them, and the seven indicators worth steering by, each with its unit and denominator.

What ESG Metrics Are and Why They Are Tracked

An ESG metric is a defined measurement of a company's environmental, social or governance performance, with a unit, a reporting boundary and a period. That definition is stricter than it sounds: without a stated boundary and denominator, a figure is a number, not a metric, and it cannot be compared across years or companies.

Metrics are tracked for three reasons that rarely arrive together. Regulation asks for specific data points, most of them in the European Sustainability Reporting Standards. Investors, lenders and insurers ask for a smaller set they can model, mostly emissions, energy and governance. Management needs a handful it can actually influence within a budget cycle. The same metric usually serves all three, which is why the list below is shorter than most ESG dashboards.

Quantitative and qualitative ESG metrics

The standards distinguish numeric data points from narrative ones, and the distinction decides how much work a metric is. Quantitative metrics resolve to a number in a defined unit: tonnes of CO2e, megawatt hours, cubic metres, a percentage. They are comparable, auditable and expensive to produce, because each one needs a data source and a boundary.

Qualitative metrics are narrative disclosures with a required content structure rather than a value: the description of a transition plan, of due diligence processes, or of how a policy is implemented and monitored. They are cheap to draft and easy to get wrong, because assurance providers test them against evidence, not against wording. Both are mandatory data points where they apply. The practical rule is that a qualitative disclosure without a quantitative anchor somewhere in it reads as a statement of intent.

ESG Metrics List and Framework Crosswalk

Most metric lists on the web name indicators without saying who asks for them or how they are calculated. The table below does both. It gives the unit or formula and the equivalent data point in the four reporting systems companies are usually asked to satisfy at once: the ESRS for EU reporting, GRI for voluntary sustainability reports, SASB for sector-specific investor disclosure, and IFRS S2 for climate reporting under the ISSB standards.

Which standard asks for which metric

The words standard, framework and questionnaire get used interchangeably and they do different things. A standard prescribes data points and how to calculate them; the ESRS and the GRI Standards work this way, and so do the ISSB standards IFRS S1 and S2. A framework prescribes a structure for what to talk about rather than the numbers themselves, which is how the TCFD recommendations operated before IFRS S2 absorbed them into a standard. A questionnaire is a buyer's or rater's own list, sent to you, mapped loosely to the standards and never identical to them.

In practice one metric set feeds all of them. Emissions calculated once to the GHG Protocol answer ESRS E1-6, GRI 305, IFRS S2 and the emissions questions in any customer ESG questionnaire. What differs is aggregation, boundary and the year requested. Building the crosswalk once, at the level of the data point rather than the report, is the difference between one measurement exercise a year and four.

The 7 Most Important ESG Metrics to Track

These seven are the ones investors, lenders and regulators request by name. Each is given with the denominator that decides whether the number is usable.

1. Greenhouse gas emissions, Scope 1, 2 and 3. Reported in tonnes of CO2e, gross, with Scope 2 given both market based and location based, and offsets disclosed separately rather than deducted. Scope 3 is where the work sits: fifteen categories, of which usually three or four dominate. My rule is to calculate the dominant categories with supplier or activity data and to keep spend-based factors only where the category is immaterial, because a spend-based total moves with inflation and prices rather than with anything you did. Anyone starting on the value chain should read the Scope 3 categories before choosing a method.

2. Emissions intensity. Tonnes of CO2e per million euro of net revenue, or per tonne of product where output is homogeneous. The absolute figure tells stakeholders about exposure, the intensity figure tells them whether decarbonisation or a bad sales year produced the improvement. Report both, on the same boundary, or the pair is not readable.

3. Energy consumption and renewable share. Total consumption in MWh, split into fuels, purchased electricity, heat and steam, with the renewable share as renewable MWh over total MWh. Two traps recur: self-generated solar that is counted in both generation and consumption, and green tariffs claimed as renewable without a contractual instrument behind them.

4. Water and waste. Water in cubic metres, with withdrawal and consumption separated because they diverge sharply in cooling-intensive sites. Waste in tonnes by hazardous and non-hazardous, with a diversion rate of diverted tonnes over total tonnes. The recurring error is taking waste tonnages from disposal invoices, which record containers collected, not mass.

5. Workforce diversity and pay gap. Share of women at board, management and total workforce level, plus the unadjusted pay gap as a percentage of the male average. The unadjusted figure is what the standards ask for. An adjusted figure that controls for role and seniority is useful for management and is not a substitute for the reported number.

6. Health and safety. Recordable injuries per million hours worked, with hours worked as the denominator rather than headcount, so that part-time and contractor exposure lands in the right place. Track near-miss reporting alongside it: a falling injury rate with a falling near-miss rate usually means reporting has stopped, not that the site has become safer.

7. Governance: board independence and ESG-linked pay. Share of independent non-executive directors, and the share of variable executive remuneration tied to sustainability targets, with the targets named. This is the metric block where wording substitutes for measurement most often. A remuneration disclosure that mentions ESG without naming the target, the weighting and the payout is a narrative, and readers of governance metrics treat it as one. Governance also carries the conduct metrics, from whistleblowing cases to supply chain due diligence coverage.

What 1,401 CSRD Reports Actually Disclose

Metric lists describe what should be reported. What is actually reported is a different question, and it is the one worth asking before committing to a metric set. My own corpus for this is 1,401 extracted European CSRD and ESRS reports, classified by NACE sector and size class, which makes it possible to see which data points companies really publish rather than which ones the standards require.

The most useful finding so far concerns emissions completeness. Across the 1,401 public 2024 and 2025 reports in the corpus, 69 percent contain complete Scope 1, 2 and 3 data, and 13 percent publish no extractable Scope figures at all. In 8 percent of the reports that do give a Scope 3 number, Scope 3 comes out smaller than Scope 1 or 2, which is a methodological red flag rather than a genuine result, since Scope 3 usually dominates the total by a wide margin.

Two lessons follow for anyone choosing metrics. First, Scope 3 is no longer the gap it is assumed to be, so a report without it now stands out. Second, being present in a data point is not the same as being comparable in it: much of the published data lacks the boundary and denominator statements that would let a reader put two companies side by side. That is where a modest report still beats a long one, and it is the cheapest quality gain available in a first reporting year.

Choosing and Implementing Your Metric Set

The selection question is not which metrics exist but how many you can produce to a defensible standard. A double materiality assessment decides the scope, and the metric set follows from it rather than from a template. This sequence works:

• Start from material topics, not from the standard. Run the materiality assessment first, then pull only the data points attached to the topics that survived it.
• Fix boundary and denominator before collecting. Which legal entities and sites are in, which unit, which denominator. Changing this later invalidates the comparison you were building.
• Set the baseline year with the data you can restate. A baseline you cannot recalculate blocks every target you attach to it.
• Name an owner per metric. Not a department, a person with access to the source system.
• Collect once, aggregate many times. One data point at source, feeding report, questionnaire and lender request.
• Promote five to ten metrics to KPI status. Each with a target, an owner and a review cycle. The remainder is documented, not managed.

The failure modes are consistent. Tracking too many metrics at once produces a dashboard nobody reviews. Keeping ESG data outside the systems that hold financial and HR data guarantees a reconciliation problem at assurance. Changing methodology between periods without restating the prior year destroys the trend, which is the only part of the data an investor reads. And targets set without a restatable baseline cannot be reported against.

Smaller companies are not exempt from being asked, only from reporting under CSRD. Following the VSME standard keeps the answer proportionate while remaining mappable to what large customers request, and the reporting process is the same at a smaller scale.

Frequently Asked Questions