Key Takeaways
Treat project failure as the default, not a freak exception
The numbers are sobering. Flyvbjerg’s database covers more than 16,000 projects across over 20 fields and 136 countries. Only 8.5 percent met both their cost and schedule targets, and just 0.5 percent also delivered promised benefits. The pattern, which he calls the Iron Law of Megaprojects, is that projects tend to exceed budgets and schedules while underdelivering benefits. It is a statistical tendency, not an unbreakable law.
Disaster has a long tail. Many project types have fat-tailed risk: most outcomes may be manageable, but extreme overruns occur far more often than a bell-curve intuition suggests. In the data, 18 percent of IT projects exceeded cost estimates by more than 50 percent, and those projects averaged 447 percent over budget. A modest contingency cannot cover every extreme.
The dataset changes the question from whether a project might fail to what makes failure so likely. That is a valuable shift from individual blame toward systems and incentives. Still, base rates are warnings, not verdicts: performance varies by project type, and the figures depend on how costs and estimates are measured. The practical lesson is not to abandon ambition, but to make evidence about comparable projects part of the approval process, before political momentum makes bad assumptions difficult to challenge.
Slow planning buys fast execution and a shorter window for disaster
Think before exposure. Planning is comparatively cheap: teams can explore designs, test assumptions, and revise documents before major spending begins. Delivery is costlier and more vulnerable to accidents, changing conditions, politics, and markets. A long project leaves more time for trouble to arrive, a vulnerability Flyvbjerg calls the window of doom.
The Empire State Building shows the alternative. Its team studied construction details closely, used proven methods, and prepared extensively before work began. The building opened on schedule in 1931 and cost about 17 percent less than its estimate. By contrast, California’s high-speed rail effort began construction without a complete, tested plan. Its scope and costs changed repeatedly, while the promised connection between Los Angeles and San Francisco remained out of reach. Starting quickly is not the same as finishing quickly.
This principle resembles the distinction in safety engineering between preparation and operations: errors found in a model or rehearsal are usually cheaper than errors found in the live system. It also challenges the common managerial habit of treating visible activity as progress. There is a real limit, however: planning can become an expensive ritual if it produces reports without resolving decisions. Its value comes from reducing uncertainty through tests and choices, not from accumulating paperwork.
Delay commitment until alternatives and risks have survived scrutiny
The first answer can trap a project. During the rush to build the Pentagon in 1941, General Brehon Somervell’s team approved a plan for a large, misshapen building on a site that would obstruct the view from Arlington National Cemetery. A nearby location better suited the project, but officials had not seriously compared alternatives. The plan was eventually moved, largely because critics and President Roosevelt intervened.
Premature lock-in is the commitment fallacy. A shallow plan is approved, activity begins, and the project becomes harder to stop even when better options emerge. Sometimes estimates are deliberately understated to secure approval, a practice called strategic misrepresentation. Later, sunk costs make cancellation feel like wasting money, even though money already spent cannot be recovered. A useful early commitment is to investigate before choosing, not to defend the first decision.
The Pentagon story makes clear that bad decisions are not always made by incompetent people. A compressed timetable, hierarchy, and the desire to show momentum can prevent capable people from asking basic questions. The same dynamic appears in organizations that reward launch announcements more than accurate diagnosis. A formal review of alternatives can help, but only when reviewers have authority to stop or reshape a project. Scrutiny without decision power is just theater.
Define the outcome before falling in love with its proposed solution
Start with the purpose. “Renovate the kitchen” describes an action, not the reason for taking it. David and Deborah began with that limited scope, then discovered structural problems and added further renovations one by one. A $170,000 estimate grew to roughly $800,000, and a three-month absence became eighteen months. Had they first discussed what they wanted from their home, they could have considered the full scope, a smaller project, or no renovation.
Gehry asked why. When officials in Bilbao proposed renovating an old warehouse into a Guggenheim museum, architect Frank Gehry asked what the project needed to achieve. The goal was to revive a struggling city by attracting visitors. He proposed a new museum on a prominent riverfront site instead. Nearly four million people visited in its first three years. Thinking from right to left means defining the desired result first, then choosing the means.
Separating ends from means is a basic discipline of decision analysis, but it is easy to lose once a team starts discussing a tangible solution. A bridge, building, or software platform can become the goal simply because it dominates the conversation. Beginning with the intended outcome creates room for less obvious options. The method works best when goals are measurable enough to guide choices, while remaining open to revision as evidence reveals what beneficiaries actually need.
Test a rough version before expensive delivery begins
Planning can be active. Flyvbjerg calls rigorous, iterative testing Pixar planning. Pixar directors turn a script into a crude film using storyboards, temporary voices, and simple sound. Audiences and colleagues review it, and the team revises the story repeatedly before full production. A typical film passes through roughly eight such rounds. The rough model makes weaknesses visible while changes are still relatively cheap.
Gehry used the same logic. For the Guggenheim Bilbao, his team moved from sketches and cardboard models to digital simulations that tested how design choices affected structure, function, and cost. Iteration can improve creativity as well as reliability: Pete Docter’s early version of Inside Out included too many emotion characters, which confused viewers, so he simplified it. This process also counters the illusion of explanatory depth, the mistaken confidence that comes from assuming a complicated idea is understood without having to explain or test it.
A prototype is more than a miniature: it is a question made tangible. A storyboard tests whether a narrative works; a model tests whether a building can be built. This is closely related to scientific experimentation, where claims become more useful when they can be exposed to evidence. The key constraint is choosing a safe test. A rough app can reach users early, but a medical device or passenger aircraft cannot responsibly treat the public as its testing ground.
Borrow experience instead of paying to be first
Experience belongs in people and tools. An experienced leader develops practical judgment, or phronesis, through repeated work in a field. Technology also carries accumulated experience: a proven design embodies lessons from earlier attempts. Yet projects often sideline both by favoring local politics, novelty, or superlatives. Seattle’s decision to build the world’s largest tunnel required an unprecedented boring machine; after only about 1,000 feet, it failed and took two years and $143 million to recover.
Being first is not automatically an advantage. Research cited in the book compared 500 brands across 50 product categories: nearly half of pioneer companies failed, compared with 8 percent of later entrants. Successful pioneers averaged 10 percent market share, while settlers averaged 28 percent. Olympic hosts face a related problem, repeatedly assembling inexperienced teams, a pattern called Eternal Beginner Syndrome. Reuse sound designs and experienced teams where possible, and test novelty more rigorously when it is essential.
Experience is not mere seniority. It is pattern recognition earned in a domain where people can learn from feedback, a distinction that helps explain why expert intuition can outperform an inexperienced person’s elaborate checklist. But prior success can also become complacency when conditions change. The strongest approach pairs experienced judgment with fresh testing: expertise proposes what deserves attention, while evidence checks whether the familiar solution still fits this site, team, and purpose.
Do not mistake celebrated rescues for the normal odds of success
A great comeback makes a great story. Jimi Hendrix’s Electric Lady recording studio began with little planning, uncertain funding, and a designer who had never built a studio. The team improvised solutions, and the finished place became influential. The troubled production of Jaws also produced creative changes that helped make the film a hit. These are examples of Albert Hirschman’s “Hiding Hand” idea: people underestimate difficulties, then discover unexpected ingenuity when problems arise.
The broader record points the other way. Flyvbjerg analyzed 2,062 projects comparable to those Hirschman studied. Four out of five exceeded costs while delivering fewer benefits than expected. Only about one in five had benefits that exceeded the cost overrun. Successful rescues are memorable partly because failed projects disappear from view. Creativity can emerge under pressure, but planning should give it room to work before delivery, when experimentation is less costly and stress is lower.
The Hiding Hand argument captures something real: people do not know every capacity they will discover while doing difficult work. But a collection of triumphant anecdotes is especially vulnerable to survivorship bias, because disasters rarely become beloved cultural stories. The distinction is between leaving room for invention and relying on desperation to produce it. The former can be designed into a project; the latter transfers risk to workers, funders, and communities who may not share in the eventual rewards.
Forecast from comparable projects, not your project's best-case story
Use the outside view. Reference-class forecasting estimates cost or time from the actual outcomes of comparable completed projects, then adjusts only when strong evidence supports a difference. The “reference class” is the broader group to which a project belongs, such as kitchen renovations, not an elaborate category tailored to make one project seem exceptional. This method counters optimism and poor anchoring by grounding estimates in experience rather than a plausible-sounding plan.
Robert Caro learned the cost of a bad anchor. He expected his biography of Robert Moses to take about a year, based on his experience writing newspaper articles. It took seven years. A forecast based on comparable books would have given him a more relevant starting point. Reference-class forecasting has improved accuracy by 30 percentage points or more for over half the project types studied. Historical results also include surprises that planners could not identify in advance.
Forecasts are often treated as neutral technical outputs, but they shape which projects receive funding and what workers are expected to deliver. The outside view helps separate a forecast from a sales pitch. Its effectiveness depends on honest comparison: a class that is too broad may obscure meaningful differences, while one tailored too tightly can smuggle optimism back in. When good data are scarce, a small collection of comparable completed projects is still more informative than confidence alone.
Cut catastrophic risk by changing exposure, not merely padding budgets
Averages cannot describe every danger. In fat-tailed project types, rare extreme overruns are more common and more severe than a normal distribution would predict. A standard contingency may cover ordinary variation but not a runaway project. Black swan management means identifying risks that could produce extreme losses and reducing their likelihood or impact, rather than relying only on a single forecast or a very large budget reserve.
The remedy is often practical. For British high-speed rail, archaeology was a foreseeable source of stoppages. Keeping qualified archaeologists available could prevent discoveries at multiple sites from cascading into delays. At the Great Chicago Fire Festival, organizers planned carefully to prevent a fire spreading, but had no backup when the ignition system failed. A second ignition method could have addressed a different, obvious failure mode. Risk reviews must ask how the project can fail in its category, not only how this specific plan might fail.
This approach reflects a central principle of resilience engineering: preventing one hazard is not enough if another failure can disable the whole operation. It also usefully distinguishes uncertainty that can be priced from vulnerability that can be designed away. Mitigation is never complete, and some safeguards cost more than the risk they reduce. The discipline is to identify plausible failure chains, prioritize the ones with severe consequences, and choose protections proportionate to the stakes.
Turn contractors into one team by sharing purpose and incentives
Heathrow Terminal 5 needed coordination. Thousands of workers from many companies had to build a complex terminal while the airport remained open. The British Airports Authority set an unusually clear deadline and used detailed digital models and factory-made components. It also changed contracts so companies could cooperate rather than protect themselves from one another’s costs. Bonuses rewarded progress against shared benchmarks.
Identity became part of delivery. Workers were encouraged to see themselves as members of the T5 project, not merely their individual firms. Managers invested in safe, well-equipped worksites, listened to workers, and made it acceptable to raise concerns. This psychological safety helped bad news travel quickly, when it could still be addressed. T5 opened on schedule and within budget. The result contrasts with Wembley Stadium, where conflict and work stoppages contributed to delay and a doubling of estimated costs.
The T5 example shows that coordination is not achieved by drawing an organization chart and hoping people cooperate. Contract design, material support, and shared identity all shape whether people can solve problems together. This resembles the logic of collective action: participants contribute more readily when the benefits and burdens are not sharply divided. A strong team culture should not suppress dissent, however. The project needs loyalty to the shared purpose and permission to challenge decisions that threaten it.
Scale giant ambitions by repeating modules that already work
Big outcomes can be assembled from small units. Modularity means building a larger system from repeatable components. In Nepal, the school program used a small number of classroom designs to build about 20,000 schools and classrooms. Completed on budget and eight years ahead of schedule, the program could learn from each batch before building the next. Repetition both improves the design and helps teams move up a positive learning curve, making later units faster and more reliable.
The approach matters for energy. Solar panels and wind turbines can be installed and connected in increments, while a conventional nuclear plant must be completed as one enormous system before producing electricity. In the book’s project data, solar, wind, roads, electricity transmission, and fossil thermal power were not fat-tailed. Flyvbjerg argues that modular renewable energy is especially relevant to climate goals, which require rapid growth in wind and solar capacity. The key design question is: what is the smallest useful unit that can be repeated?
Modularity reduces the size of each bet: a defect in one component need not invalidate the entire investment, and learning can travel from unit to unit. It also shifts construction toward controlled manufacturing and assembly, where repetition is easier to standardize. Not every system can be divided cleanly. Interfaces between modules can create their own complexity, and some projects depend on tightly coupled components. The task is to modularize where independence is genuine, not to mistake a parts list for a scalable system.
Analysis
The book’s strongest contribution is a change in the unit of attention. Project failure is usually narrated as a late-stage management problem: a contractor stumbled, a machine broke, or a leader failed to react. Flyvbjerg instead traces breakdowns upstream, to the choice of goal, the quality of estimates, the timing of commitment, and the structure of the project itself. That shift matters because it changes what leaders can still control. Once concrete is poured or a public promise is made, correction becomes expensive and politically difficult.
The argument combines three kinds of discipline. Cognitive discipline means resisting optimism, anchoring, and the belief that one’s project is exceptional. Organizational discipline means facing incentives that reward approval and visible starts more than realistic estimates. Design discipline means building with iteration, experience, and repeatable components. These are mutually reinforcing: modular designs create more opportunities to learn, while better forecasts make it easier to fund the time and preparation that learning requires.
The evidence is powerful, but its application needs care. Base rates depend on project category, measurement choices, and historical conditions. A reference class is not useful if it conceals a decisive difference, and modularity is not a universal answer where components are tightly interdependent. Nor can competent process remove political conflict or guarantee public benefit. The method is best understood as risk reduction, not certainty.
That restraint strengthens the book’s central claim. A successful project is not simply a bold vision plus heroic effort. It is a chain of decisions that protects the vision from both human bias and structural fragility. The most practical test is whether a team has learned enough before delivery, chosen a goal worth pursuing, and made the inevitable surprises survivable. In that sense, disciplined planning is not the enemy of ambition. It is what keeps ambition from consuming its own resources before it can deliver anything useful.
Review Summary
How Big Things Get Done receives high praise for its data-driven insights into successful project management. Readers appreciate Flyvbjerg's emphasis on thorough planning, modular approaches, and risk mitigation. The book's practical advice, backed by extensive research, resonates with professionals across industries. Critics note some redundancy and oversimplification, but most find the content valuable for both large-scale projects and personal endeavors. The book's accessible style and compelling examples make it a recommended read for anyone involved in project planning and execution.
People Also Read
Glossary
Iron Law of Megaprojects
Typical project underperformanceFlyvbjerg’s name for the recurring statistical pattern in large projects: costs exceed estimates, schedules run late, and delivered benefits fall short. It is not a universal physical law. The claim describes a high and persistent probability of underperformance across many project types, based on the author’s database.
Think slow, act fast
Plan carefully, deliver swiftlyThe book’s central project principle. Teams should use planning to explore, test, and resolve problems before costly delivery begins. Thorough preparation can make execution smoother and shorter, while reducing the time a project remains exposed to shocks, delays, and expensive changes.
Commitment fallacy
Premature lock-inThe tendency to settle too early on a project’s purpose, design, site, or solution, before alternatives and risks have been seriously examined. Early commitment can create lock-in: people and organizations continue as if they have no choice, even when evidence suggests a better path.
Pixar planning
Iterative planning through testsA planning approach that uses repeated, relatively inexpensive simulations, models, or rough versions to discover and correct weaknesses before full production or construction. Pixar’s storyboards and trial screenings illustrate the method. Each cycle generates feedback, allowing the plan to improve before changes become costly.
Phronesis
Experience-based practical wisdomAn Aristotelian concept used in the book for practical wisdom: the judgment needed to recognize what is good in a specific situation and to act effectively. It grows through experience, including tacit knowledge that people may be unable to express fully in rules or instructions.
Uniqueness bias
Seeing projects as exceptionalThe tendency to treat a project as so unusual that lessons from comparable projects seem irrelevant. This bias can lead planners to ignore relevant experience, underuse proven methods, and forecast from an optimistic story about their own project rather than from the record of similar work.
Reference-class forecasting
Forecasting from comparable outcomesA forecasting method that identifies a broad class of similar completed projects, uses their actual costs, schedules, or benefits as an empirical anchor, and adjusts only when evidence supports doing so. It counters optimistic planning and captures the effects of surprises that may not be individually predictable.
Regression to the tail
Extreme outcomes keep appearingFlyvbjerg’s term for the behavior of some fat-tailed project outcomes, where new extremes can appear often enough that the average is unstable or misleading. In these settings, planners should not assume that results will cluster near the mean. They need to examine the distribution and reduce exposure to extreme losses.
Black swan management
Mitigating extreme project risksAn approach to project risk that studies severe outcomes in comparable projects, identifies the mechanisms that produced them, and takes steps to prevent or limit those mechanisms. Rather than treating extreme losses as unknowable accidents, it seeks to reduce exposure through practical safeguards, contingency plans, and faster delivery.
Eternal Beginner Syndrome
Repeatedly rebuilding without learningA recurring condition in projects such as the Olympic Games, where each host organization starts with a largely inexperienced team and does not systematically carry knowledge forward. Frequent turnover and political preferences for local contracts can prevent the project system from benefiting from accumulated experience.
Scale-free scalability
Scaling by repeating the same patternThe capacity to enlarge or reduce a system by combining repeatable units while preserving its basic operating pattern. The book uses examples such as server farms, solar arrays, and modular schools. This structure can support incremental investment, learning, and expansion without requiring one indivisible project.
FAQ
What's How Big Things Get Done about?
- Focus on Megaprojects: The book delves into the complexities of managing large-scale projects, known as megaprojects, which often encounter cost overruns and delays.
- Success vs. Failure: It contrasts successful projects, like the Guggenheim Museum Bilbao, with failures, such as the Sydney Opera House, to identify patterns and lessons.
- Universal Principles: The insights apply not only to megaprojects but also to smaller projects, making the book broadly relevant.
Why should I read How Big Things Get Done?
- Practical Insights: Offers actionable strategies for improving project management, valuable for professionals across various fields.
- Data-Driven Analysis: Backed by research and a database of over 16,000 projects, providing empirical evidence for its claims.
- Avoiding Common Pitfalls: Teaches readers how to avoid common mistakes in project planning and execution, enhancing success chances.
What are the key takeaways of How Big Things Get Done?
- Think Slow, Act Fast: Advocates for thorough planning before execution, leading to smoother and faster delivery.
- Reference-Class Forecasting: Uses data from similar past projects for accurate cost and time estimates, reducing overruns.
- Team Dynamics Matter: Emphasizes the importance of a cohesive team with a common purpose for project success.
What is the "Iron Law of Megaprojects" in How Big Things Get Done?
- Common Outcomes: States that megaprojects are "over budget, over time, under benefits, over and over again."
- Statistical Evidence: Research shows only 8.5% of projects meet cost and time estimates, while 99.5% exceed budget or schedule.
- Implications for Planning: Helps project managers anticipate challenges and implement risk mitigation strategies.
How does How Big Things Get Done define "Pixar Planning"?
- Iterative Process: Emphasizes experimentation and iteration in project development for continuous improvement.
- Creative Collaboration: Involves diverse teams collaborating to test and refine ideas, similar to Pixar's film development.
- Focus on Learning: Encourages learning from both failures and successes to ensure well-tested final products.
What is "Reference-Class Forecasting" in How Big Things Get Done?
- Data-Driven Estimates: Involves using historical data from similar projects for realistic estimates, anchoring forecasts in reality.
- Avoiding Uniqueness Bias: Counters the tendency to view projects as unique, which can lead to overly optimistic forecasts.
- Practical Application: Improves accuracy of cost and time estimates, leading to better project outcomes.
What is the "commitment fallacy" in How Big Things Get Done?
- Definition: Refers to the tendency to continue investing in failing projects due to prior investments.
- Importance of Flexibility: Suggests "committing to not committing" for reassessment and pivoting when projects falter.
- Real-World Implications: Provides examples of projects suffering from this fallacy, highlighting the need to recognize when to abandon failing initiatives.
How does How Big Things Get Done address the role of psychology in project management?
- Optimism Bias: Discusses how optimism can lead to unrealistic expectations and poor planning, resulting in failures.
- Cognitive Biases: Highlights biases like the planning fallacy and uniqueness bias that skew project forecasts.
- Awareness and Mitigation: Understanding these factors allows managers to mitigate their impact for realistic planning.
How does How Big Things Get Done define modularity?
- Modularity Explained: Describes constructing large projects from smaller, interchangeable components for flexibility.
- Examples of Modularity: Uses Lego blocks analogy to illustrate combining small units for larger structures.
- Benefits of Modularity: Projects can be completed faster, with lower costs and reduced risks, adapting to changes.
What are the best quotes from How Big Things Get Done and what do they mean?
- "Big is best built from small.": Emphasizes modularity in project management, suggesting smaller parts lead to success.
- "Repetition is the mother of learning.": Highlights iterative processes, where repeated actions lead to improved performance.
- "Commit to not committing.": Encourages flexibility and willingness to abandon failing projects for better resource management.
How can I apply the lessons from How Big Things Get Done to my projects?
- Adopt a Modular Approach: Identify basic building blocks of your project to break down complex tasks into manageable units.
- Implement Iterative Processes: Create cycles of planning, execution, and evaluation to learn and improve outcomes.
- Be Willing to Pivot: Regularly assess project viability and be prepared to change direction if necessary.
How does How Big Things Get Done address the climate crisis?
- Urgency of Action: Emphasizes rapid and effective project delivery to combat climate change.
- Modular Solutions: Suggests modular approaches for quick deployment of renewable energy projects.
- Call to Action: Encourages innovative strategies prioritizing sustainability for meaningful climate action.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.