Search, Sprint, Ramp Up: How Large Capital Project Decisions Actually Unfold

Across the world's major and emerging economies alike, investment in infrastructure and industrial capacity has become one of the defining features of economic policy. Highways, metro systems, airports, ports, refineries, power plants, LNG terminals and new manufacturing complexes are being planned and approved at a scale few generations have seen. Each of these ventures absorbs capital for a decade or more, ties up scarce engineering talent, and shapes the economic life of the regions around it for many decades afterwards. The stakes of large capital project decisions are therefore not only financial. They are social, political and environmental, and they reach far beyond the project boundary. When such a decision is made well, it creates lasting value. When it is made poorly, the consequences are felt by owners, lenders, governments, communities and taxpayers for years.

This comprehensive blog will explore how large capital project decisions actually unfold over time, from the first search for a viable concept, through the intense sprint of engineering and construction, to the often underestimated period of ramp-up. We will delve into why the most important choices are made long before a project formally exists, and why the front end of a project so often goes unfunded and unmanaged. Furthermore, we will examine the role of bias and of front-end discipline, the different phase systems used by sponsors and owners, the concept of governability, and the trap of escalating commitment once money is spent. Whether you are a project sponsor, owner's representative, government official, investor or project development professional, understanding how these decisions unfold is key to delivering projects that are not only built, but worth building.

1. The project you approve isn't the one you start

A large capital project can be delivered "successfully" by every conventional measure and still fail. The plant may be completed close to budget, the commissioning may be smooth and the safety record may be excellent, and yet within a few years the asset may be written down. This happens when the market it was designed to serve, the permits it relied upon or the partners who were expected to supply it have shifted beneath it. In such cases the project has not been built badly. It has been built well, but it was the wrong project for the world it had to live in.

This distinction is central to understanding large capital projects. The steel and concrete are the visible part of the venture. What determines the outcome is the web of institutions, partners, regulators and communities that surrounds it. A project that does not fit its social, regulatory and environmental setting will struggle regardless of the quality of its engineering.

Studies of large capital projects make for sobering reading in this respect. Such projects tend not to land in the middle ground. Most turn out to be either clear successes or clear failures, with relatively few that are merely somewhat late or somewhat over budget. Researchers who have examined hundreds of these ventures conclude that sponsors could have prevented nearly every failure. The implication is significant: failure on large capital projects is mostly not a matter of bad luck. It is a matter of decisions.

When the causes of failure are examined closely, poor estimating is rarely the root cause. The more powerful drivers are externalities that nobody priced, a thin understanding of the institutions on which the project depends, and the behaviour of the coalition of sponsors, lenders, governments and contractors once conditions become difficult. Estimating matters, but it sits downstream of these factors.

When a board approves a large capital project, therefore, it is not simply approving a design. It is approving a bet on a coalition and on an environment. The project that is approved is seldom the project that is eventually started.

2. Three phases, one clock

A useful way to understand this journey is to view a large capital project as passing through three phases on a single sponsor's clock: search, sprint and ramp-up.

• Search is the long front end, in which the sponsor looks for a viable concept, builds the coalition, secures rights and permits, and shapes the deal.
• Sprint is the engineering and construction phase, in which commitments have been made and the task is to deliver.
• Ramp-up is the period from first operation to stable performance, in which it becomes clear whether the asset will actually generate the expected returns.

A major international study of sixty large capital projects, treated as a set of compared cases rather than a statistical sample, found that the full journey from the start of search to revenue took about ten years. The average project in that set cost close to one billion dollars, at roughly $985 million.

Separate research on large capital projects places the beginning of "shaping" some six to twelve years before the investment is formally authorised. Because that research draws on a different sample and uses a different starting point, the two figures should not be added together. Both, however, point to the same conclusion: the clock starts long before anyone calls the venture a project.

Figure 1: Search, sprint and ramp-up run on one sponsor's clock, roughly ten years from the start of search to revenue.

This is arguably the single most important idea in project leadership. The decisions that matter most are made before there is a project manager, a schedule or a budget line. By the time a project becomes "real", much of its fate has already been determined.

3. Search: the long front end nobody budgets for

In the study of sixty projects, the search phase averaged 79 months, which is more than six years of work before the sprint begins. Yet most organisations have no budget line, no team structure and no leadership model for those years. Search tends to happen in the gaps: within a business development group, through a consultant, or with engineers on part-time loan from other assignments.

The financial picture is similar. On standard projects, front-end spending typically runs at about 3–5% of total cost. On complex projects it can rise to as much as 35%. Many executives regard this spending as frivolous, seeing studies, presentations and legal work as a distraction before the "real work" begins. The question of why so much money should be spent before a decision has even been taken is heard frequently in boardrooms.

The answer is that front-end money is best understood as risk capital. It is the least expensive money that will ever be spent on the project, because it buys information at a time when changes cost very little. Reducing it does not save money. It simply moves the cost into the sprint, where every surprise is paid for in materials, rework and delay.

One of the most neglected elements of the front end is the Basic Data: the technical facts on which the design rests. These include site conditions, soil and seabed surveys, feedstock composition, water chemistry and climate. When Basic Data is missing, late or wrong, errors cascade. The design is built on it, the estimate is built on the design, and the schedule is built on the estimate. A frequently cited example is an offshore project in which the seafloor had never been properly surveyed, so that everything built on that missing knowledge had to be redone. No quality of later project management can recover a foundation that was never laid.

For sponsor organisations, the lesson is clear. Search should be treated as real work, with a real budget, dedicated people and clear accountability.

4. Shaping: same word, two different jobs

"Shaping" is among the most frequently used terms in front-end work, and among the most frequently misunderstood. In practice it refers to two distinct tasks, and teams that use the word in different senses can lose months as a result.

The first meaning is shaping the coalition. Over the years of search, a sponsor brings together partners, governments, lenders, communities and regulators, and moulds the venture so that enough of them want it to succeed. The most capable sponsors treat this as a competitive capability that improves across successive projects, rather than as a one-off negotiation. It is slow, political work, and it cannot be outsourced.

The second meaning is shaping the deal: dividing the value of the project fairly enough that all partners will agree to a scope that can then be frozen. Some of the most expensive mistakes on large capital projects occur here. One is the sponsor who seeks to retain all of the value, leaving partners and host governments with too little reason to remain committed. Another is the assumption that the deal can be worked out later, so that engineering proceeds while the commercial basis remains unsettled. A deal that is not settled will be reopened, and usually at the worst possible moment.

A telling case is a project worth about seven billion dollars that spent roughly a quarter of a billion before anyone discovered that it had no secure feedstock. The engineering was advancing, but the coalition was not. The part of the deal that mattered most had never been shaped.

The two meanings should therefore be kept separate, and both should be used. Coalition shaping concerns who wants the project and why. Deal shaping concerns who receives what, so that the scope can be frozen. Blending the two conceals the gaps in each.

5. Thinking under turbulence and ambiguity

Large capital projects are not neat decision trees. It is not possible to list every branch, assign a probability to each and roll the result up into a clean expected value. The environment surrounding a ten-year venture is turbulent: governments change, commodity cycles turn, technologies evolve and partners merge.

Leading sponsors therefore rely on scenarios and simulations far more than on a single probabilistic run. They ask what worlds the project may have to operate in before they ask for a P50 cost. When probabilities cannot honestly be assigned, as is usually the case early in search, scenarios judged on plausibility rather than likelihood help to reframe the problem. They reveal which assumptions the whole business case rests upon, and which of those assumptions would break it.

There is also good reason for caution about the way quantitative risk analysis is often practised. A Monte Carlo model built on an immature scope, with ranges supplied by the same team that wants the project approved, can do real harm. It produces a precise-looking number that offers false comfort. The difficulty lies not in the mathematics but in applying a tool designed for well-defined uncertainty to a problem that remains ambiguous.

The practical lesson is to match the tool to the phase. In the early stages, scenarios should be used to explore and frame the problem. Later, once the scope and Basic Data are firm, quantitative risk analysis begins to earn its place.

6. The inside view: why every search looks shorter and cheaper than it will be

Almost every project team believes its project is different, with better people, a smarter contracting strategy and a clearer plan than those that came before. This belief is natural, and it is one of the most reliable sources of failure in the industry.

When people plan, they tend to produce plans that sit close to the best case. They imagine the steps proceeding roughly as intended and add a modest margin. This tendency, known as the planning fallacy, affects experts as much as anyone else. The remedy is surprisingly simple: before trusting an internal estimate, examine how similar projects have actually turned out.

Two further forces compound the problem. The first is uniqueness bias, the belief that a project is so special that the experience of others does not apply. The second is less innocent: deliberate misrepresentation, in which costs are shaded down and benefits shaded up in order to secure approval. Both are present on most large capital projects, and they often appear in the same individuals at the same time.

The tool that addresses both is reference-class forecasting. It involves identifying a class of comparable past projects, examining the actual distribution of their outcomes and placing the new project within that distribution. Only then are adjustments made for what is genuinely different. The process is humbling by design.

7. The argument: is it bias or bad front-end work?

There is a genuine debate among those who study large capital projects, and practitioners benefit from understanding it rather than adopting a position by habit.

One school of thought places the primary blame on psychology and power. Optimism, the planning fallacy and strategic misrepresentation result in projects being approved on numbers they could never achieve. On this view, improving forecasting and incentives would resolve most of the problem.

The other school challenges this account. If bias explained everything, it asks, how can one account for the fact that more than a third of projects in large capital project databases turn out to be outstanding successes, delivered by people subject to the same biases? On this view the real cause of failure is the quality of front-end work: Basic Data, scope definition, the team and the deal. Bias is present, but it is not the principal lever.

There is common ground between the two. Researchers on the psychological side have long accepted that honest bias and deliberate misrepresentation operate together. There is also a third perspective that neither school fully captures: institutions and governability, meaning how well the coalition and its rules hold up when the world changes.

A balanced reading suggests that all three matter. Bias explains why front-end work is skipped or rushed. Weak front-end work explains why that bias turns into failure. Weak institutions explain why the failure cannot be recovered. Effective sponsors work on all three.

8. Sprint: start only when the package is committable

The sprint, comprising detailed engineering, procurement and construction, averaged 49 months in the study of sixty projects. This is a substantial period, yet it is shorter than search, and it should feel different. Search is concerned with exploration and keeping options open. Sprint is concerned with commitment and pace.

The critical question at the transition is whether the package is committable. This means asking whether the coalition is settled, whether the rights and permits are secure, and whether risk has been allocated in a way the parties will actually honour. If the answer is no, starting the sprint does not accelerate the project. It locks in its problems.

A helpful principle here is to think slow and act fast. Planning should be slow, careful and detailed, while changes are inexpensive. Delivery should then be rapid, because every month of delay costs real money and increases exposure to surprise. Failed projects tend to reverse this pattern, rushing the thinking in order to "get going" and then moving slowly through delivery as unresolved issues return one after another.

Two sponsor behaviours frequently trigger this reversal. The first is schedule pressure from the top, the demand to have the project now. It appears decisive, but it usually means that the front end is cut short. The second is the last-minute cost cut, such as a demand to remove 20% from the estimate just before authorisation. Neither changes the true cost of the work, and both can initiate a cascade: thinner scope, weaker contingency, contractor claims, rework and, ultimately, a larger overrun than the cut was intended to prevent.

9. Different phase systems: the sponsor's clock, owner gates and governance stages

Different phase systems are often confused in practice, and it is therefore worth setting them side by side.

The search–sprint–ramp-up clock describes the sponsor's journey through time and coalition-building. It offers a view of the entire life of the venture.

The front-end loading (FEL) gate system is an owner's discipline for the front end, structured around three gates. FEL-1 builds the business case. FEL-2 selects the concept and freezes the scope. FEL-3 completes definition up to the final investment decision. The evidence for doing this well is strong: in large capital project databases, projects with the best front-end loading averaged more than four times the net present value per dollar invested of those with the weakest.

A third approach, emerging from governance research, organises large capital projects into three stages: Forecasting, Organizing and Executing. It places outside-view forecasting at the beginning, followed by the design of organisation and incentives, with execution coming only after both.

These are best understood as three different clocks, each built for a different purpose. They can be used together, but their phase names should not be treated as interchangeable. Search is not FEL-1, and sprint is not Executing. When a steering committee confuses them, it is often a sign that a step is being skipped.

Figure 2: Three phase systems built for different purposes: map them side by side, but do not merge their phase names.

10. Governability: holding the coalition together when surprise arrives

Every large capital project encounters surprise. A government falls, a key partner is acquired, a commodity price halves, or an unforeseen geotechnical problem emerges. Such events cannot be scripted. What can be done is to design the project so that it is able to absorb them.

This property may be described as governability: the ability of a project's coalition to hold together and act jointly when the unexpected occurs. In practice it rests on a small number of arrangements agreed early, while commitments are still inexpensive:

• Reserves, both financial and in schedule, that are genuine rather than merely a line in the estimate.
• Triggers for joint action, meaning agreed conditions under which the parties come together to renegotiate rather than retreating to their individual contracts.
• Flexibility built into scope, phasing and commercial terms, so that the project can bend without breaking.
• Risk allocation by capability, so that each risk sits with the party best able to control or carry it.

The final point is where considerable damage is often done. A common sponsor belief is that contractors should carry the risk, and so the owner presses for a lump-sum price on the assumption that the problem has been solved. A lump-sum price, however, is a floor rather than a ceiling. When a contractor cannot absorb a risk it was never able to control, the cost returns through claims, disputes, slowdowns and sometimes insolvency. The risk has not been transferred. It has been concealed until it is at its most expensive.

It is also helpful to separate two ideas that are frequently merged. Governability concerns how the coalition behaves under surprise. Governance concerns the owner's internal rules: who decides go or no-go at each gate, on what evidence and with what degree of independence. A project needs both, and strength in one does not compensate for weakness in the other.

11. Ramp-up and the trap of escalation

Ramp-up is the phase in which the truth emerges. Every assumption made during search and every shortcut taken during the sprint becomes visible in the operating results.

Studies of large capital projects show that ramp-up typically ends in one of three ways. The project performs, reaching its design output and economics. It limps on as a lame duck, operating but never delivering what was promised. Or it undergoes financial restructuring, in which the original owners absorb the loss and another party acquires the asset at a discount.

It is important to recognise that being ready to operate is not the same as operating well. Mechanical completion is a milestone, not an outcome. What counts as acceptable performance also depends heavily on the type of asset. An oil production project running at 70% of planned output in its early years may be tolerable. A liquefied natural gas plant at 70% is a serious failure, because its economics depend on operating close to full capacity from the outset.

Ramp-up is also where one of the most dangerous decision traps closes in: escalation of commitment. When a project struggles, there is a strong temptation to continue investing because of what has already been spent, on the grounds that the organisation has come too far to stop. Yet money already spent is gone regardless of what is decided next. The only relevant question is whether the next unit of investment is worth making. People and organisations, particularly those who championed the original decision, find this very difficult to accept, and the stronger their sense of responsibility for that decision, the harder it becomes.

A project delivered on time is not necessarily a project worth building. Some of the soundest decisions on large ventures are those to stop, scale back or restructure early, and they are rarely celebrated.

12. What to do on Monday

For those who lead or sponsor large capital projects, several practical steps follow from the discussion above.

Fund search as real work. The front end deserves a budget, a team and a leader, and front-end spending should be treated as risk capital rather than overhead.

Run a premortem before the decision becomes formal. The team is asked to imagine that it is some years in the future and the project has failed, and each member then writes down the reasons why. This is one of the least expensive and most effective ways to surface concerns that would not be raised in a conventional review.

Add an outside-view check to every gate. Before accepting the team's estimate, it should be compared with how similar projects have actually performed, by identifying the relevant reference class and the project's position within its distribution.

Secure the Basic Data before freezing scope. No concept selection or scope freeze should proceed on assumed site, feedstock or environmental data.

Write the reserve and renegotiation rules while commitments are still inexpensive. Partners should agree early on what happens when circumstances change: who contributes, who decides and what triggers a joint review.

Avoid punishing project directors for honest overruns. Where every overrun becomes a career event, project directors learn to conceal money in the estimate and the organisation ceases to receive accurate information. Early bad news is valuable, and those who bring it should be rewarded.

Know which clock the project is on. Leaders should recognise whether a project is in search, sprint or ramp-up, and lead accordingly: exploring during search, committing during the sprint and remaining candid during ramp-up.

Closing

The central lesson of large capital projects is that most of the decision is made before anyone breaks ground. The choices made years before mobilisation, concerning the coalition, the deal, the data and the allocation of risk, shape everything that follows. The sprint is where a project becomes visible, but the search is where it is won or lost.

For sponsors, owners, governments and investors seeking better outcomes, leadership attention should be directed to where the leverage lies: at the front end, and well before it feels urgent.

---

References

• Flyvbjerg, B. (2021). Top ten behavioral biases in project management: An overview. Project Management Journal, 52(6), 531–546. https://doi.org/10.1177/87569728211049046
• Flyvbjerg, B., & Gardner, D. (2023). How Big Things Get Done: The Surprising Factors That Determine the Fate of Every Project, from Home Renovations to Space Exploration and Everything in Between. New York: Currency.
• Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
• Klein, G. (2007). Performing a project premortem. Harvard Business Review, 85(9), 18–19.
• Lovallo, D., Cristofaro, M., & Flyvbjerg, B. (2023). Governing large projects: A three-stage process to get it right. Academy of Management Perspectives, 37(2), 1–19. https://doi.org/10.5465/amp.2021.0129
• Merrow, E. W. (2024). Industrial Megaprojects: Concepts, Strategies, and Practices for Success (2nd ed.). Hoboken, NJ: Wiley.
• Miller, R., & Lessard, D. R. (2000). The Strategic Management of Large Engineering Projects: Shaping Institutions, Risks, and Governance. Cambridge, MA: MIT Press.
• Ramírez, R., & Wilkinson, A. (2016). Strategic Reframing: The Oxford Scenario Planning Approach. Oxford: Oxford University Press.
• Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance, 16(1), 27–44.