Upstream capital projects carry unique complexities other megaprojects don't: The value potential depends on a reservoir that cannot be fully understood, even after millions to billions of dollars of investment and decades of production. This uncertainty, along with a range of technical and commercial complexities, dictates whether a project succeeds. Estimating the value potential is expensive and happens well before sanction in an appraisal that characterizes the reservoir and a concept selection phase that weighs genuine alternatives and challenges underlying assumptions. Subsurface uncertainty is the biggest risk to value, and it runs in both directions. When teams underestimate uncertainty or frame known risks as manageable to secure approval, the result is underdelivery and loss, made worse by large-scale complexity, new technology, and premature concept lock-in. But the same uncertainty can surprise on the upside: Some projects find better rock than sanctioned, only to leave that value on the table because the concept was locked in too narrowly to capture it. Either way, the chosen concept sets the ceiling on value.

The stakes attached to that concept choice are rising. The oil and gas sector faces a continued need for reserves amid increasing project complexity and growing shareholder expectations. More than a decade of underinvestment in exploration has thinned the opportunity hopper, and too few investable projects are coming through to replace what's being produced. That scarcity is dangerous. When the pipeline is thin, operators are tempted to "make it work" by advancing marginal concepts, downplaying uncertainty, and forcing projects toward sanction that a robust hopper would have filtered out. Learning from experience and taking an objective, risk-balanced approach to development decisions is more important than ever.

Figure 1

Project net present values (NPVs) often diverge from sanctioned estimates (see Figure 1). No company sanctions a project expecting to lose money or erode value, yet some do. Other projects, often because of subsurface factors, exceed expectations but may leave money on the table.

A recent portfolio review reinforced these findings: Across 110 oil and gas investments, gross costs exceeded early-stage estimates by $138 billion, with 21 projects accounting for 93% of overspending.

The question is, Why do upstream projects continue to fall into these traps?

Enduring mistakes exacerbated by human cognitive biases 

These are among the most common and enduring mistakes, along with illustrative examples of why they still occur:

Mistake	Examples of why it endures
Premature concept lock-in	A senior steer becomes the answer. The project team treats a preference as a decision and won't challenge upward, so the concept is never truly reopened.
Under-appraisal	Appraisal costs time and money now to avoid losses years later. A well that only confirms the base case looks wasted in hindsight, and a thin hopper adds pressure to keep moving.
Weak, early framing	Framing is performed too early by a small, junior team, without senior or supplier challenge and with no standard for what good looks like. The project advances on the first plausible frame, not the right one.
Facilities sized to one view	Facilities must freeze to hold cost and schedule; the subsurface is resolved later. Designed to a single case instead of a range, the concept locks before the reservoir is understood.
False precision in economics	Steering committees prefer a confident number to an honest range. A wide band reads as a team that doesn't know its asset, so the process rewards the look of certainty over the uncomfortable reality.
Contracting set too early	Procurement runs on its own clock—yard slots, market windows, the urge to lock capacity. Committing before scope settles feels like control; it just moves risk to where it prices back later.

These errors are familiar, but sunflower, anchoring, and confirmation biases exacerbate them. Teams need to employ deliberate operating processes to mitigate the effect of these biases when early warnings emerge.

Unfortunately, teams often ignore warning signs and rarely self-correct. Bad projects usually get worse, and after the costs, scope, or schedule deteriorates, the project likely needs a reset. Racing to produce technology “firsts” accelerates this dynamic, particularly when compensating for subsurface complexity. Recognition of these mistakes before sanction is what separates calibrated decision making from procedural assurance.

Figure 2

Mistakes happen at different points in the project life cycle (see Figure 2). Value erosion related to execution often appears early, while subsurface and operating surprises emerge later, during ramp-up and early production. By then, the sanctioned concept is increasingly locked in, limiting self-correction and driving actual value away from the final investment decision.

Four decision strategies, one capital system

More governance is not the answer. International oil companies already have stage gates, technical assurance, peer reviews, and capital committees. Teams need to honestly challenge decisions, increase rigor, strengthen owner-team capability and bias awareness, and tailor contracting to the project’s uncertainty and financial exposure. The following framework can help teams make more disciplined decisions.

Uncertainty is multidimensional, affected by technical, commercial, fiscal, market, regulatory, and execution factors. Exposure is the magnitude of irreversible loss if the decision is suboptimal or wrong. Together, uncertainty and exposure cover four quadrants (see Figure 3), each calling for a different approach. An effective capital system must be able to apply all four.

Figure 3

In practice, each approach requires a different balance of testing, decision rigor, speed, and execution control:

• Explore boldly (high uncertainty/lower exposure): Define the cheapest test that could change the next decision, set explicit exit criteria, and recycle capital quickly. The mistake here would be overcommitting and overtesting for the uncertainty. (Applies to frontier prospects, basin openers, and technology pilots)
  
  
• Ensure decision quality (high uncertainty/higher exposure): Prioritize disciplined evaluation of alternatives, value-of-information logic, early alignment with senior decision makers, and explicit decision pathways with precommitted off-ramps. Include at least one robust strategy and one high-value alternative so leadership can choose across resilience, flexibility, phasing, and delay. (Applies to large-scale liquefied natural gas projects, multibillion-dollar concepts, and first-of-its-kind technology deployments)
  
  
• Move fast (low uncertainty/lower exposure): Industrialize through standards; delegated authority; preferred suppliers; repeatable, standardized designs; and exception-based review. The cost of waiting is often larger than the value of another assurance loop. (Applies to routine workovers, lift optimization, interventions, and incremental field optimization)
  
  
• Execute with discipline (low uncertainty/higher exposure): Treat familiar programs as capital-at-risk systems. The following factors matter most: owner-led control, cost and schedule discipline, change control, commissioning readiness, and rapid variance correction. (Applies to tight drilling pads, brownfield infill and enhanced oil recovery, and subsea tiebacks.)

The approach can’t be static. Through their life cycles, opportunities migrate across quadrants as uncertainty resolves and exposure evolves, so timing matters. A risk-balanced approach applies more rigor and senior-level involvement when necessary, without costly assurance of routine projects. Without these approaches, teams face leakage later: cost overruns, scope changes, and contracting and execution problems.

Recognizing risks 

Attrition, outsourcing, cost reduction, and portfolio churn have weakened many owner teams. These teams are often less able to distinguish project risk from contractor risk, technical uncertainty from governance noise, or concept problems from execution problems. Leaders who have full-cycle major project experience and can challenge contractors and counter biases are scarcer than the headcount suggests.

The engineering, procurement, and construction (EPC) market has seen the same project-performance data and may have learned these lessons better than owner teams. EPCs no longer absorb risks that owners can’t define. Risk-sharing works only when both sides can price risks honestly, requiring that owner teams be capable of managing these structures.

So why do projects continue to fall short?

Many causes of value erosion in upstream capital projects (premature concept lock-in, under-appraisal, weak assumption challenges, overconfidence in base cases, and insufficient attention to uncertainty and life cycle flexibility) have been well understood for years. Yet organizations continue to err because decisions are still shaped by human involvement, incentives, hierarchies, and cognitive biases (such as sunflower, confirmation, anchoring, overconfidence, and near-term reward bias).

AI and agentic decision support can help. They can surface disconfirming evidence, test assumptions, and benchmark against past outcomes. But AI is an enabler, not a fix. Bolt it onto a system that still rewards clearing the gate over getting the concept right, and it will only make biased decisions faster. Without deeper change to how capital decisions are governed, resourced, and rewarded, the value destruction stays exactly as it is today: avoidable but not avoided.

Six questions for leadership teams 

The next step for many oil and gas companies is to install an upgraded capital management system that is calibrated to risk and past performance and built on bias mitigation and rebuilt owner capability.

Answer these six questions to determine whether your organization has such a system:

1. Looking back, do we know how much NPV has been gained or lost vs. the FID case, and can we trace losses to subsurface surprise, concept choice, scope growth, schedule slippage, claims, commissioning delay, or underperformance?
2. Which current projects can be categorized as high uncertainty and high exposure, and are they receiving materially different governance, owner-team talent, and executive attention?
3. For each project in the portfolio, which uncertainties change the investment decision, and which tests merely make teams more comfortable?
4. Across the organization, what biases emerge through the process, and how should we mitigate their effect?
5. Where are EPC contracts being used to transfer risks that contractors can’t control—and therefore will price back through premiums, claims, or change orders?
6. Do you have a mechanism to test the amount of bias that your key projects could be carrying? 

Owner teams that achieve sustained performance in upstream capital deployment know when to be more deliberate, when to simplify, when to build flexibility into the infrastructure, and when to own the risks contractors can’t absorb. Leading organizations also learn from past performance and can self-identify and mitigate their biases. In a sector that must keep investing just to stand still, this capital discipline and self-awareness are what separate organizations that consistently deliver from those that keep relearning the same lessons.