A precise environmental score can create false confidence when small changes in uncertain assumptions can change which option appears best.
An executive team comparing major technologies may be presented with an apparently clean answer. Option A has the lowest environmental burden. Option B is less efficient. Option C requires a 20 per cent improvement to reach the frontier. The mathematics looks disciplined, the ranking is visible and the decision appears ready for approval.
But what happens if the inputs are not known with the same certainty as the output suggests?
Environmental decisions frequently combine measurements, estimates, databases, allocation choices and modelling assumptions. When those inputs are uncertain, the strategic issue is not simply whether one alternative has a better score. It is whether the decision remains defensible when reasonable uncertainty is allowed to move through the model.
A 2017 study by Ewertowska and colleagues combined life cycle assessment, data envelopment analysis and Monte Carlo simulation to test exactly this problem. Their electricity-generation case demonstrated that a technology classified as efficient under nominal values could become inefficient under another plausible realisation of uncertain environmental parameters. Improvement targets also changed between scenarios. The lesson for leaders is broader than the method: decision quality depends on the stability of the conclusion, not the apparent precision of the calculation.
The Strategic Context
Sustainability analysis is increasingly used to influence capital allocation, technology selection, supplier choice, portfolio prioritisation and operating-model design. These decisions can be material, long-lived and difficult to reverse. Yet the supporting evidence often contains a mixture of high-quality measurements and weaker assumptions.
Life cycle assessment is valuable because it broadens the system boundary beyond a single site or direct emission source. Data envelopment analysis can then compare multiple inputs and outputs without requiring a single subjective weighting scheme. Together they can expose alternatives that use environmental resources more efficiently relative to peers.
The difficulty is that relative efficiency calculations are only as stable as their inputs.
Ewertowska et al. focused on uncertainty originating from issues such as lack of data and inaccurate measurements. Their approach sampled uncertain environmental inputs repeatedly and recalculated efficiency. This changed the question from "What is the efficiency score?" to "How often does this option remain efficient across plausible conditions?"
That second question is much closer to the information an investment committee needs.
Related article: Relative Efficiency Can Still Be Unsustainable: The Executive Case for Environmental Budgets
What Leaders Commonly Misread
Precision is mistaken for certainty
A model may report three decimal places while the underlying inventory data carries substantial uncertainty. The numerical form of the output can make the result look more certain than the evidence warrants.
Executives should distinguish between calculation precision and decision confidence. The former is a property of the model output. The latter depends on data quality, assumption sensitivity, model structure and the stability of the recommendation.
The nominal case is treated as the truth
A nominal case is usually a reference point, not a guarantee. If an alternative is efficient only under one central estimate but loses that status across many plausible scenarios, the nominal result is strategically fragile.
This does not mean the model is useless. It means the model should reveal fragility rather than conceal it.
Rankings are assumed to be more robust than targets
The 2017 study showed that uncertainty could affect both whether a technology was classified as efficient and the reduction target assigned to an inefficient technology. That matters because organisations often convert model outputs directly into commitments, budgets and performance targets.
A target that appears to require a 26 per cent reduction in one case but a materially lower reduction in another is not merely a technical issue. It changes project scope, investment requirements and possibly the viability of the initiative.
Uncertainty is treated as a reason to delay
There is an opposite failure as well. Some organisations respond to imperfect information by postponing decisions indefinitely.
That is not disciplined uncertainty management. The executive task is to determine whether the uncertainty is decision-relevant. If all plausible cases support the same decision, more analysis may add little value. If small changes reverse the ranking, further evidence, staged investment or optionality may be justified.
Reframing the Issue
The right question is not:
Which alternative has the best environmental score?
It is:
Which alternative remains acceptable across the range of conditions that matter, and what would make us change course?
This reframing moves environmental assessment from static measurement into decision intelligence.
A robust decision does not require every variable to be known. It requires leaders to understand which uncertainties can change the choice, which uncertainties only change the size of the benefit, and which uncertainties are immaterial.
Three distinctions are useful.
Parameter uncertainty concerns uncertain values inside the model, such as environmental impacts or inventory estimates.
Structural uncertainty concerns how the model represents reality, including boundaries, causal relationships and allocation choices.
Decision uncertainty concerns whether the final recommendation changes when the first two forms of uncertainty are tested.
The third is the executive concern. It tells leadership whether uncertainty is merely analytical noise or a threat to the investment logic.
Strategic Analysis: Stability Matters More Than the Winning Score
A useful ranking should survive reasonable challenge. If a technology is the leading option in 95 per cent of plausible cases, that is a different decision from an option that leads in 51 per cent.
The point is not to establish an arbitrary confidence threshold. It is to make the distribution visible.
Consider a hypothetical manufacturing company selecting between three process technologies. One has the lowest nominal carbon burden but depends heavily on a poorly characterised upstream material. Another has a slightly higher nominal impact but far better data quality and a stable supply chain. A third has the strongest upside but requires an unproven recovery process.
A deterministic comparison may favour the first. A robustness analysis may show that the second has a lower probability of materially underperforming the environmental target. The decision then becomes a trade-off between nominal advantage and confidence.
That is a capital-allocation discussion, not merely an environmental one.
Uncertainty should influence reversibility
The more uncertain the evidence, the more valuable reversibility becomes.
If the decision can be staged, piloted or contracted with exit rights, management can preserve option value while learning. If the decision involves a large irreversible asset, the evidence threshold should be higher.
This leads to a practical rule:
The burden of proof should rise with irreversibility, downside exposure and the sensitivity of the decision to uncertain assumptions.
Data quality is an economic variable
Better data has a cost. So does bad data.
An organisation should not pursue perfect information. It should invest in information when the expected decision value exceeds the cost of obtaining it. If one uncertain parameter repeatedly changes the preferred option, improving that parameter may be worth far more than refining dozens of variables that do not affect the decision.
This is where uncertainty analysis becomes a prioritisation tool. It directs attention to the evidence that can actually change the choice.
Relative efficiency is not absolute acceptability
Even a robust relative ranking answers only one question: which option performs better compared with the alternatives or frontier used in the analysis.
It does not necessarily tell leaders whether the absolute environmental burden is sustainable. An option can be consistently "best" and still exceed an ecological limit.
That distinction becomes critical when sustainability objectives are expressed against external thresholds rather than internal improvement.
Related article: Relative Efficiency Can Still Be Unsustainable: The Executive Case for Environmental Budgets
Decision Framework
Before approving a sustainability-sensitive investment, leaders can apply a six-part robustness test.
| Test | Executive question | Decision implication |
|---|---|---|
| Boundary | What is inside and outside the assessment? | Hidden impacts may invalidate the ranking. |
| Data quality | Which inputs are measured, estimated or assumed? | Weak critical inputs deserve targeted validation. |
| Sensitivity | Which variables can materially move the result? | Focus analysis on decision-changing variables. |
| Ranking stability | Does the preferred option remain preferred across plausible cases? | Fragile rankings call for caution or optionality. |
| Downside | What happens if the optimistic assumptions fail together? | Define exposure and contingency. |
| Reversibility | Can the commitment be staged or altered? | Higher uncertainty favours reversible pathways. |
The framework is intentionally decision-oriented. It does not prescribe one statistical technique. Monte Carlo simulation is one useful method when probability distributions can be reasonably represented, but the executive discipline is broader: test whether the conclusion survives uncertainty.
From Strategy to Execution
Immediate action
Require major environmental business cases to distinguish nominal results from uncertainty-tested results. Identify the three to five assumptions most capable of changing the decision. Record whether each is measured, estimated or externally sourced.
Where the ranking is stable, avoid unnecessary analysis. Where it is unstable, determine whether better data, a pilot or a different contracting structure can reduce the exposure.
Medium-term capability building
Build uncertainty treatment into sustainability governance rather than leaving it to specialist analysts. Investment committees should be able to read ranges, distributions and sensitivities as comfortably as they read net present value scenarios.
Data governance should also classify environmental data by quality and decision criticality. Not every uncertain field deserves equal improvement effort.
Long-term strategic positioning
Organisations with mature decision intelligence will increasingly treat sustainability models as living evidence systems. As new data arrives, rankings and thresholds should be recalculated, not frozen at the point of project approval.
This matters particularly for long-lived assets, emerging technologies and supply chains where upstream conditions can change materially during the life of the investment.
Signals to Monitor
Leaders should pay attention when:
- the preferred option changes under small assumption variations;
- environmental targets move materially between scenarios;
- a critical input relies on old, proxy or geographically mismatched data;
- project teams report a single score without sensitivity information;
- new supplier, technology or regulatory information changes a high-impact parameter;
- an irreversible commitment is being justified by weakly characterised evidence;
- the decision remains unchanged despite repeated refinement, suggesting further analysis may have low value.
Questions for the Leadership Team
- Which uncertain assumption could most easily reverse the decision we are about to make?
- Are we seeing the distribution of plausible outcomes or only the nominal case?
- What is the cost of being wrong, and how reversible is the commitment?
- Which additional evidence would actually change the decision rather than simply improve the report?
- Are environmental improvement targets robust enough to be converted into funded commitments?
- Is the preferred option merely relatively efficient, or is it also acceptable against the limits that matter?
Closing Perspective
A sustainability model should reduce uncertainty about a decision, not hide uncertainty behind a precise number.
The most valuable analytical question is therefore not whether a score can be calculated more accurately. It is whether the decision remains sound when reality refuses to match the central estimate.
Leaders should demand evidence that exposes that fragility early. When the ranking is robust, they can act with greater confidence. When it is not, they can buy information, preserve reversibility or redesign the choice before capital is locked in.
Precision is useful. Decision stability is what makes precision strategically valuable.
Source References
- Ewertowska, A., Pozo, C., Gavaldá, J., Jiménez, L. & Guillén-Gosálbez, G. 2017, 'Combined use of life cycle assessment, data envelopment analysis and Monte Carlo simulation for quantifying environmental efficiencies under uncertainty', Journal of Cleaner Production, vol. 166, pp. 771-783, doi:10.1016/j.jclepro.2017.07.215.
- Wolff, A., Gondran, N. & Brodhag, C. 2017, 'Detecting unsustainable pressures exerted on biodiversity by a company. Application to the food portfolio of a retailer', Journal of Cleaner Production, vol. 166, pp. 784-797, doi:10.1016/j.jclepro.2017.08.057.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.