Why Big Projects Fail (And 5 Rules For Beating the Odds)

By: Danielle Dy Buncio, CEO

Every major building project begins with optimism. You announce the development, you unveil the concept drawings, you revel in the ambitious vision and the promise of impact.

And then… things go off the rails. According to Bent Flyvberg and his landmark book How Big Things Get Done, 99.5% of large capital projects fail to deliver on time, on budget, and on scope.

And it’s not unique to any one industry. Whether it’s an ERP rollout, a hospital expansion, a manufacturing facility, a transit line, or even the Olympics, large projects collapse under the same weight of flawed planning, overconfidence, and rigid execution.

But some owners – a small group of outliers – consistently achieve a different arc. They experience more predictable cost, schedule, and scope outcomes. And this is because they plan differently. They front-load learning. They de-risk decisions early. They use digital models to tighten the variance window before the first shovel hits the ground.

For years we’ve observed that the clients who embrace early learning, model-based decision-making, and disciplined planning consistently deliver better outcomes. Flyvbjerg’s research gives the data-driven explanation for why those practices work. And in this article we’ll look at the patterns behind these outlier results.

Rule 1: Plan by Experimenting

Planning is not a static phase – or at least it shouldn’t be. Instead, the great planners learn by doing. It’s how you find blind spots, adjust scope, and avoid mistakes.

BIM and Digital Twins allow you to simulate and scenario plan quickly. By modeling your site and layering in the actual conditions on-site, you can run cheap, iterative simulations to reduce risk and course correct as you go.

Rule 2: Spend More Time in the Cheap Phase

Flyvbjerg calls the early, low-cost stage the “safe zone.” Once contractors are on site, cranes are running, and cash is burning, the clock starts ticking. That’s when projects become fragile.

This idea isn’t new. In the 1970s, a Stanford civil engineering professor named Boyd Paulson charted the relationship between when decisions are made and their impact on cost and schedule. The concept later became widely known in the industry through what many call the MacLeamy Curve, though the underlying insight traces back to Paulson’s work.

The curve makes it clear that the earlier you make decisions, the cheaper they are to change. The later you make them, the more expensive (and risky) they become.

Virtual Design & Construction was created to take advantage of that curve. By modeling conditions and resolving conflicts early, owners are able to move much more of the decision-making into that low-cost phase.

The longer you stay in the low-cost planning zone, the shorter and safer the high-cost execution zone will be. Digital models, clash detection and AI-assisted schedule can help, compressing your window of high-risk vulnerability.

Rule 3: Expect Black Swans

Nassim Nicholas Taleb calls a black swan a “once-in-a-million” event. A labor strike. A supplier collapse. A global pandemic.

He also argues that while you can’t anticipate what type of black swan you might encounter, it’s a safe bet that every project will encounter some type of black swan.

The smartest planners assume this. they build flexibility into their contracts and their models, assuming that some type of disruption will be inevitable.

And they work to close the vulnerability window as quickly as possible. The longer a project stays in its high-cost, high-exposure execution phase, the more chances there are for a black swan to fly through and derail it. The shorter that window becomes, the lower the odds of a major shock.

That’s where integrated data and digital twins help. When you can stitch design, schedule, and cost data together into a single source of truth, you can run scenarios and stress-test where you might run into trouble.

While you might not be able to predict every disruption, you can reduce the time you’re exposed to them.

Rule 4: Build with Lego Blocks

The Empire State Building was completed in just over a year. And a major reason was because every floor was basically identical. They didn’t get fancy with layouts, and opted for a consistent “Lego block” (although they almost certainly called it something different back then. Were Lincoln Logs a thing yet?)

Limiting variables reduces the number of failure points. And modular design is one way to limit variables. Each step of the process becomes easier. These days, prefabrication and digital workflows can create this type of repeatability at scale.

Rule 5. Borrow from the Reference Class

When estimating cost and schedule, we’re wired for optimism. We can’t help ourselves.

Flyvbjerg’s suggests the answer lies in using “reference class forecasting”. You base your estimates on the actual performance of similar past projects. This has a track record of outperforming any other type of bottom-up modeling.

This means using historical performance data from your portfolio to inform new project baselines. Something data-rich BIM and digital twin environments now make possible.

Learn and Experiment. Don’t Gamble.

Most projects fail because teams don’t get enough chances to learn early, when changes are cheap and risk is low. But the owners who consistently beat the odds work differently.

They use VDC as a planning discipline. They bring design, field conditions, schedule, and cost into one environment, to make faster decisions with better information. Every clash resolved, every sequence validated is one less surprise during execution.

You no longer need to rely on optimism and gut feel. You now have access to the tools and technologies to experiment early, back your intuition up with data, and anticipate problems before they happen.

And in the coming years, that’s how smart organizations will really get big things done.