When ESG KPI's Fail in High-Risk Industries and Drive the Wrong Behavior
A sustainability KPI can look clean in a board pack and still make operations worse.
That is the uncomfortable truth in high-risk, capital-intensive industries. Oil and gas, construction, mining, utilities, marine operations, manufacturing, and major infrastructure projects all rely on metrics to control cost, risk, safety, emissions, waste, water, local content, and contractor performance. The problem is not measurement itself. The problem is measurement that rewards the wrong behavior.
Poorly designed ESG KPIs can encourage teams to game numbers, suppress bad news, move impacts outside the reporting boundary, chase targets with little operational value, and report improvement that does not reflect real performance. In these sectors, that is not a reporting inconvenience. It can distort investment decisions, procurement choices, project execution, community relations, and regulatory risk.
The better question is not, “Do we have ESG metrics?” Most companies do. The better question is, “What behavior do these metrics create?”
Bad sustainability KPIs create commercial risk
ESG measurement has matured. Investors, lenders, regulators, customers, and communities expect credible data. IFRS S1 and IFRS S2, issued by the International Sustainability Standards Board, focus on sustainability-related financial disclosures and climate-related risks and opportunities that could affect enterprise value. The GRI Standards focus on impacts on the economy, environment, and people. The GHG Protocol provides widely used accounting rules for greenhouse gas emissions across scopes and boundaries. ISO standards such as ISO 14001 and ISO 14064 support environmental management and emissions quantification. IFC Performance Standards and World Bank Group Environmental, Health, and Safety Guidelines shape expectations for many financed projects.
These frameworks do not solve KPI behavior on their own. They provide structure, principles, and disclosure expectations. Companies still decide what to measure, how to define it, who owns it, how it affects incentives, and how much assurance it receives.
That is where things often go wrong.
A metric becomes risky when it is easy to improve the number without improving the underlying condition. In project-based and asset-heavy industries, that can happen fast because teams work under intense pressure. Production targets, liquidated damages, permit limits, outage windows, vessel availability, contractor claims, commodity prices, and community commitments all compete for attention. If a sustainability KPI sits inside bonus plans or contract scorecards, people will respond to it.
They may respond in useful ways. They may reduce fuel use, improve waste segregation, inspect erosion controls, repair leaks early, or buy from credible local suppliers.
They may also respond in unintended ways.
Common failure modes include:
Gaming
Teams change behavior to satisfy the metric while avoiding the intent.
Selective reporting
Good data moves upward. Bad data stays in spreadsheets, field logs, or contractor files.
Boundary manipulation
Impacts move to subcontractors, suppliers, disposal vendors, joint ventures, temporary sites, or downstream users.
Data suppression
Near misses, spills, grievances, water exceedances, or waste nonconformances are under-recorded because reporting them hurts the score.
Short-term optimization
Teams hit an annual KPI by delaying maintenance, deferring cleanup, changing shipment timing, or pushing impacts into the next reporting period.
Target chasing
Managers pursue visible KPI movement instead of material risk reduction.
These behaviors are not theoretical. They align with well-established management research on performance measurement, including Goodhart’s law, often summarized as: when a measure becomes a target, it can stop being a good measure. The same principle appears in safety, education, finance, public administration, and environmental management. The stronger the incentive, the greater the risk of distortion.
In ESG governance, that means you should test a metric like a control. It should not only describe performance. It should help prevent predictable misuse.
The metric type matters, but no single type is enough
Many KPI debates fail because teams mix different measure types and treat them as interchangeable. Activity metrics, leading indicators, lagging indicators, and outcome-based measures all have value. They also have blind spots.
Measure type | What it tells you | Where it helps | Where it can mislead |
Activity metrics | What was done | Training delivered, audits completed, inspections performed, supplier assessments issued | Can reward volume without quality or impact |
Leading indicators | Whether controls and behaviors are in place before harm occurs | Preventive maintenance, leak detection, waste segregation checks, permit-to-work quality, closure of corrective actions | Can become box-ticking if not linked to outcomes |
Lagging indicators | What already happened | Emissions released, injuries, spills, complaints, waste disposed, water withdrawn | Can arrive too late and may encourage underreporting |
Outcome-based measures | Whether the intended condition improved | Lower absolute emissions, improved water quality, restored habitat, reduced exposure, durable local economic benefit | Can be influenced by external factors and may take years to show change |
Activity metrics are attractive because they are easy to count. An EPC contractor can report how many environmental inspections were completed. A utility can report the number of workforce hours trained on spill prevention. A marine operator can count vessel energy audits. These numbers show effort. They do not prove that risk fell.
Leading indicators are more useful when they test the health of controls. For example, a refinery may track the percentage of high-priority methane leak repairs completed within defined time limits. A construction project may track whether sediment controls remain effective after rainfall, not only whether they were installed. A mining operation may track closure of critical tailings governance actions, not only meetings held.
Lagging indicators remain necessary. Absolute emissions, recordable injuries, spills, noncompliances, water withdrawals, waste disposed, and community grievances matter. They tell management what happened. They also carry underreporting risk when attached to incentives. Safety professionals have long recognized that injury rates can fall because work became safer, or because reporting became weaker. The same logic applies to ESG performance.
Outcome-based indicators are the most meaningful and often the hardest to design. They ask whether the condition changed. Did ecosystem quality improve after restoration? Did local procurement build durable supplier capability, or did it only reroute spend through nominally local intermediaries? Did water stewardship reduce basin stress, or did the facility only reduce its own intake while the supply chain increased consumption?
No single measure type is sufficient. A credible KPI system uses a small set of linked measures. The activity confirms work was done. The leading indicator tests whether controls are working. The lagging indicator records actual consequences. The outcome measure tests whether the sustainability objective was achieved.
If those measures point in different directions, leadership should pay attention. A site that reports excellent inspection completion but rising spills has a control problem. A project that reports falling emissions intensity but rising absolute emissions has a strategy problem. A contractor with perfect self-reported local content but repeated community complaints has a credibility problem.
Where ESG KPIs distort decisions
Weak sustainability KPIs rarely fail in isolation. They shape trade-offs. In heavy industry, those trade-offs involve capital, schedule, contractors, permits, and operating risk.
Emissions metrics can hide growth in absolute impact
Carbon intensity metrics can be useful. They show emissions per unit of production, revenue, kilometer moved, barrel produced, ton processed, megawatt-hour generated, or square foot built. They help compare efficiency across assets and time.
They can also hide rising absolute emissions.
A liquefied natural gas facility may reduce emissions per ton produced while total emissions rise due to expansion. A mining company may improve haulage efficiency per ton of ore moved while lower ore grades require more material movement. A data center or utility may reduce emissions intensity while total electricity demand grows. A contractor may lower fuel use per work hour while total equipment hours increase due to rework.
The GHG Protocol separates organizational boundaries, operational boundaries, and scope categories for a reason. Boundary choice affects what gets counted. If a KPI focuses only on Scope 1 fuel combustion at owned sites, it may encourage outsourcing to contractors without reducing total emissions. If procurement teams receive credit for lower reported embedded carbon but rely on supplier declarations with weak data quality, the number may improve before the underlying supply base changes.
Better emissions KPIs pair intensity with absolute measures and boundary clarity. They also distinguish between reductions from operational change, asset divestment, grid decarbonization, production shifts, offsets, and accounting adjustments. IFRS S2 and related sustainability disclosure practice have increased scrutiny of transition plans, assumptions, and climate-related metrics. That direction matters because investors need to understand whether performance reflects real decarbonization or portfolio reshuffling.
Waste KPIs can reward diversion on paper
Waste diversion rates are common in construction, infrastructure, manufacturing, and shutdown projects. They can encourage recycling and reuse. They can also reward classification games.
A project may improve its diversion rate by sending mixed material to a facility that claims recovery, even when end markets are uncertain. A demolition contractor may report high recycling volumes while paying less attention to hazardous fractions. A manufacturing site may reclassify byproducts as beneficial use without tracking actual downstream fate. A marine operation may reduce onboard waste landing at one port by shifting disposal to another port with weaker oversight.
The KPI improves. The waste risk may not.
A better waste KPI looks beyond the first transfer. It may track waste prevention, hazardous waste reduction, contamination rates, verified end destination, chain-of-custody quality, and nonconforming loads. For circularity claims, teams need evidence that material re-enters productive use. GRI waste disclosures and many regulatory systems emphasize waste generation, composition, recovery, disposal methods, and hazardous waste controls. Those categories help, but only if you test definitions and vendors.
Water KPIs can ignore basin reality
Water intensity is useful inside a plant, mine, or power asset. It can also distract from local water stress. Saving one gallon in a water-abundant area does not carry the same risk relevance as saving one gallon in a water-stressed basin. A site may report lower freshwater withdrawal per unit while total withdrawal strains a local aquifer during drought. A contractor may reduce potable water use by switching to alternative sources that affect other users.
The World Bank, IFC, and other finance institutions place strong emphasis on context, stakeholder impacts, and environmental and social risk management. That context matters in water. A good KPI should distinguish source, quality, timing, discharge condition, reuse, and local basin stress. Outcome-based indicators may include compliance with discharge limits, reduced dependence on stressed sources, improved water-balance reliability, or verified community water-access commitments.
Biodiversity metrics can become acreage accounting
Biodiversity KPIs often rely on area disturbed, area restored, trees planted, habitat offset hectares, or survey completion. These are visible and easy to count. They can also miss ecological quality.
A pipeline project can reduce reported disturbance width while fragmenting sensitive habitat. A mine can report hectares rehabilitated while vegetation structure, species composition, soil function, and long-term viability remain poor. An infrastructure project can plant trees that fail after two dry seasons. A port expansion can complete all required surveys yet still face cumulative impacts on marine species.
IFC Performance Standard 6, the World Bank Group EHS Guidelines, and emerging nature-related reporting practices all point toward materiality, mitigation hierarchy, critical habitat, ecosystem services, and monitoring. A KPI that counts activity without ecological condition will not manage biodiversity risk. Better indicators combine avoidance and minimization decisions, habitat quality, survival rates, invasive species control, and independent ecological monitoring over time.
Local content and social value can reward pass-through arrangements
Local content targets can support jobs, supplier development, community acceptance, and license to operate. They can also create paper compliance.
A project may count spend through a locally registered entity that imports most labor, equipment, and materials. A contractor may hire locally for low-skill, temporary roles while keeping specialized, higher-value work external. A supplier may meet ownership criteria but lack real operational capability. Training hours may rise while long-term employment does not.
OECD guidance on responsible business conduct and IFC standards both emphasize meaningful stakeholder engagement, labor practices, and local impacts. For local content, that means measuring more than spend. Better KPIs may include verified local wages paid, supplier development milestones, retention after training, subcontract value retained locally, payment timeliness for smaller suppliers, and grievance trends linked to employment and procurement.
The commercial risk is real. Inflated local content claims can damage trust faster than missing a target with a credible explanation.
Procurement KPIs can push impacts into the supply chain
Procurement teams often receive sustainability targets for supplier screening, recycled content, certified materials, low-carbon products, human rights due diligence, and local spend. These targets can improve ESG governance. They can also shift impacts outside the reporting boundary.
If buyers are rewarded for lowest reported carbon per unit, suppliers may use generic databases, selective product declarations, or narrow life-cycle boundaries. If procurement KPIs focus on certification status only, teams may prefer suppliers with paperwork over suppliers with stronger practices but less mature documentation. If contract scoring rewards low waste at the main site, suppliers may overpackage offsite or absorb scrap elsewhere.
The GHG Protocol Scope 3 Standard highlights the complexity of value-chain emissions. ISO life-cycle and product carbon footprint standards also show why assumptions, allocation methods, and boundaries matter. Procurement KPIs need data quality scoring, supplier assurance, and category-specific rules. High-risk categories deserve more scrutiny than low-risk spend.
Safety and contractor KPIs can suppress bad news
Safety is often treated separately from ESG, but it belongs in any serious ESG performance discussion for high-risk sectors. Contractor safety metrics can create some of the strongest perverse incentives.
A lost-time injury rate target can encourage modified duties that avoid classification rather than better hazard control. A zero-spill KPI can discourage reporting of small spills. A near-miss target can fail in two ways. If teams demand a minimum number, reports may become low-quality. If teams punish high numbers, reporting dries up.
Professional safety practice has moved toward balancing lagging indicators with leading indicators such as critical control verification, high-potential event learning, leadership field engagement quality, and corrective action closure. The same principle applies across sustainability reporting. If a metric makes bad news costly, bad news will travel slowly.
Why KPI design fails in practice
Most weak KPIs fail for practical reasons, not intellectual ones. The people designing them often understand the sustainability issue. They underestimate how the metric will behave under pressure.
Denominators are vague or too convenient
Many ESG metrics rely on denominators: per ton, per barrel, per mile, per project hour, per revenue dollar, per vessel day, per unit produced, per contract value. The denominator can change the story.
Revenue-based intensity can improve because of price changes, not better environmental performance. Production-based intensity can improve if low-performing assets shut down during the period. Project-hour denominators can shift when subcontractor hours are excluded. Vessel-day metrics can look better if idle time is treated differently. Per-square-foot construction metrics can hide complex ground conditions, design changes, or tenant specifications.
The denominator must match the decision the KPI is meant to guide. If leadership needs to understand total environmental pressure, absolute values matter. If operations need efficiency comparisons, intensity helps. If investors need transition risk, they need both, with clear explanations.
Targets are detached from operational reality
A target that ignores project phase, asset age, geology, weather, grid availability, permitting constraints, technology readiness, or supply-chain maturity invites workaround behavior.
An EPC project in early earthworks will have different waste, dust, diesel, and water profiles than a commissioning phase. A brownfield refinery cannot follow the same emissions pathway as a new greenfield facility without capital plans and outage windows. A mining operation in a remote region may face limited recycling outlets and long logistics routes. A marine contractor may depend on vessel class, charter terms, port fuel availability, and weather windows.
Ambitious targets have value. Detached targets create fiction. Teams respond by renaming activities, changing boundaries, delaying bad news, or negotiating exceptions after the fact.
Better targets account for baseline quality, operational constraints, capital requirements, and control ownership. They still stretch performance, but they do not pretend the asset is something it is not.
Incentives rely on self-reported data
Self-reporting is unavoidable in large operations. It is not sufficient for incentive-grade metrics.
If contractor bonuses depend on self-reported emissions reduction, local content, incident rates, waste diversion, or training completion, the data needs verification. Assurance does not have to mean full external assurance for every field. It does mean risk-based checks, traceable records, clear definitions, sample testing, and consequences for misreporting.
The International Auditing and Assurance Standards Board, ISO standards, and sustainability assurance practice all emphasize evidence, criteria, and assurance boundaries. The detail varies by engagement, but the principle is stable: credible reporting needs suitable criteria and sufficient evidence.
A contractor scorecard that affects payment should not rely only on unchecked monthly spreadsheets.
Boundaries encourage impact transfer
Boundary manipulation is one of the most common failures in ESG KPIs. A company lowers its reported impact by moving the activity elsewhere.
Examples are common in high-risk industries:
An operator sells a high-emitting asset and reports lower Scope 1 emissions, while the asset continues operating under new ownership.
A constructor reduces onsite waste by prefabricating more components, but supplier scrap increases and is not counted.
A utility reduces direct workforce safety exposure by outsourcing high-risk maintenance, while contractor risk rises.
A port operator improves its own energy KPI while tenants and vessels account for most local emissions.
A mining company excludes temporary construction camps from water and waste metrics, even though camps drive local impacts.
This does not mean outsourcing, divestment, or prefabrication are bad. They may create real benefits. The KPI should show whether the impact fell or moved.
The GHG Protocol’s treatment of scopes is useful here because it forces a broader view of direct and value-chain emissions. GRI’s impact focus also helps because it asks organizations to consider significant impacts connected to their activities and business relationships. Good KPI design borrows from both views.
Better KPI design starts with behavior
A strong sustainability KPI system starts with the decision it needs to improve. Then it tests how the metric might be misused.
This is where experienced leadership adds value. The finance team may ask whether the metric is auditable. The project team may ask whether it reflects site reality. The commercial team may ask whether contract terms reward the right actions. The sustainability team may ask whether the indicator reflects material impacts. Operations may ask whether people can control it.
All of those questions matter.
A better design process includes several practical steps.
Define the purpose before the number
A KPI should have a clear job. Is it for disclosure, capital allocation, operational control, contractor management, lender reporting, regulatory compliance, or executive pay? One metric rarely serves all purposes well.
A board-level climate KPI may track absolute emissions reduction against a transition plan. A site-level emissions KPI may track fuel efficiency, flaring events, methane leak repair, or equipment idle time. A contractor KPI may track verified fuel data and adherence to low-emission work methods. These measures connect but aren't identical.
Use paired indicators where gaming risk is high
Single metrics are easier to game. Paired indicators make distortion more visible.
For example:
Weak signal alone | Better paired signal |
Emissions intensity | Absolute emissions, production context, and capital plan progress |
Waste diversion rate | Waste generated, hazardous waste, contamination, and verified end destination |
Local procurement spend | Value retained locally, supplier capability, payment terms, and job quality |
Training hours | Competency checks, field behavior, and incident learning |
Zero spills | Spill reporting quality, response time, root-cause closure, and high-risk observations |
The goal is not to create a dashboard with dozens of measures. The goal is to prevent one clean number from hiding a messy reality.
Build data quality into the KPI
Data quality should not sit in a footnote. It should affect confidence in the reported result.
For high-impact KPIs, companies can rate data by source, method, completeness, timeliness, and assurance status. Metered fuel data is different from estimated fuel use. Supplier-specific emissions data is different from a generic factor. Verified waste tickets are different from contractor summary claims. Community grievance records are different from informal anecdotal updates.
A KPI that improves while data quality worsens should trigger challenge, not applause.
Assign ownership to people who can act
Ownership often fails when sustainability teams are made accountable for outcomes controlled by engineering, procurement, operations, contractors, or joint venture partners.
If procurement controls supplier requirements, procurement should own relevant supplier KPIs. If construction controls waste segregation and subcontractor methods, construction should own those controls. If asset management controls maintenance and capital spend, it should own emissions-reduction actions tied to equipment. Sustainability should support methods, governance, reporting, challenge, and assurance. It should not be the sole owner of operational performance it cannot direct.
Clear ownership also reduces target chasing. Teams know which decisions they must change.
Review KPIs when the operating model changes
A KPI that worked during design may fail during construction. A construction KPI may fail during commissioning. An operating asset KPI may fail after acquisition, divestment, technology change, or new regulation.
Periodic review is not a weakness. It is part of control. IFRS, GRI, ISO management system approaches, and responsible business conduct guidance all reinforce the need for governance, continual improvement, and relevant information. In practice, that means stress-testing KPIs at least annually and after major operational changes.
The review should ask:
Does the KPI still reflect a material issue?
Can teams improve the number without improving performance?
Has the boundary changed?
Has data quality improved or weakened?
Are incentives causing underreporting or delay?
Do leading indicators and outcome measures tell a consistent story?
Should the metric be retired, revised, or assured differently?
The leadership test for KPI integrity
Senior leaders do not need to approve every calculation method. They do need to understand whether sustainability KPIs are creating the right pressure.
The most useful questions are direct:
What behavior is this KPI likely to drive in the field?
Could a project manager, contractor, supplier, or asset team game it?
Does reported improvement reflect real performance, or a change in boundary, denominator, timing, or data source?
Are we rewarding low reported incidents, or better risk control and stronger reporting?
Are we reducing total impact, or moving it to contractors, suppliers, communities, future years, or other owners?
Does the KPI balance leading indicators with lagging and outcome-based indicators?
Is the data good enough for the decision or incentive attached to it?
Who owns the actions needed to improve the result?
What would we learn if the metric moved in the wrong direction?
Should incentives be redesigned to reward transparency, control quality, and durable outcomes?
These questions matter most when ESG metrics affect pay, contract awards, preferred supplier status, project approvals, or public claims. Incentives raise the stakes. They also raise the likelihood of defensive reporting.
A commercially aware KPI design does not punish teams for surfacing problems. It rewards early identification, credible data, prompt correction, and verified improvement. That distinction matters. In high-risk industries, silence rarely signals control.
The most mature organizations treat KPI design as part of ESG governance, not as a reporting exercise. They connect sustainability reporting to operational risk management, internal controls, procurement governance, project assurance, and capital planning. They accept that some metrics need external assurance. Others need internal audit. Others need field verification by technical specialists.
They also avoid false precision. A sustainability KPI doesn't become reliable just because it has two decimal places. It becomes reliable when the method is clear, the boundary is honest, the data is tested, and the behavior it creates is understood.
Final professional takeaway: judge a sustainability KPI not only by what it measures, but by the behavior it creates. If it rewards gaming, silence, boundary shifting, or short-term target chasing, it is not a performance tool. It is a risk disguised as a metric.
