What 22,000 Projects Reveal About Success and Failure

Here's a number that should keep every executive awake at night: out of more than 22,000 projects studied across every sector, only about 0.5% delivered on budget, on time, and on the benefits they promised. One project in 200 to 250.

Alexander Budzier, Fellow in Management Practice at Oxford's Saïd Business School, shared that statistic on a recent episode of The Shift Code Podcast with host Pierre Le Manh, and it’s just the starting point of a conversation that challenges everything we think we know about project success.

The Sydney Opera House paradox

Consider the following numbers:

• 50% of projects deliver on budget or better
• 8% on budget and on time
• 0.5% on budget, time, and benefits

If they depress you, Alexander offers a fascinating counterpoint. Consider the Sydney Opera House: massively delayed, absurdly over budget, and so painful it cost architect Jørn Utzon his career. By every traditional metric, a catastrophe. Yet today it is one of the world’s most recognizable buildings, credited with helping redefine Sydney and generate lasting economic value.

The reverse is equally instructive. The German supermarket giant spent half a billion euros on an accounting system that went live on time and won industry awards. Fifteen months later, a new CEO pulled the plug and wrote off the entire investment. The system didn’t fit the business.

As Budzier puts it, a project can be a project management success and still be a total business failure, and vice versa. That's why the iron triangle was never the whole story.

The biggest red flag you're not tracking

So what actually predicts disaster? Alexander’s research uncovered something surprising. The single biggest red flag for a cost blowout is when people involved in the project start saying “this project is unique”.

He calls it the uniqueness trap. The moment a team believes their project has no precedent, they stop learning from history, dismiss reference data, and convince themselves that normal rules don’t apply. That’s when budgets spiral. Other warning signs are equally telling but rarely tracked: critical roles left unfilled and failure to involve end users in IT projects.

These weak signals are often missing from steering committee dashboards, but they matter far more than the metrics that do.

Think slow, act fast

Most projects spend roughly 3-5% of total cost on planning. The projects that actually deliver  on time, budget, and benefits invest between 20% and 30%. That’s not a typo. The best performers spend roughly four to ten times more on getting the plan right before a line of code is written, or a shovel hits the ground.

Alexander points to the Madrid Metro, where the team discovered you don’t accelerate a project by cutting planning short. You accelerate by thinking harder upfront, then executing with precision. He’s not anti-agile, but he pushes back hard on the interpretation that says ‘just start building.’ Whether you’re vibe-coding an app or constructing a nuclear facility, skipping front-end thinking creates debt that eventually comes due.

What leaders must unlearn

What do project leaders need to unlearn?

First, unlearn the idea that delivering on time and on budget means everything is fine. The real question is about long-term value. What did your project actually contribute to the organization, the profession, and the project’s long-term legacy?

Second, the harder one, unlearn doing the work yourself. Most project leaders rose through specialist ranks. They got promoted because they were brilliant at their work. But once you’re leading, your job is to get the work done, not to do it. That difference is enormous.

A brilliant leader trapped in a broken system will still crash. The most impactful thing a leader can do is build the system itself so the project thrives regardless of any single person's heroics.

The AI Productivity paradox

Generative AI boosts individual productivity by roughly 40%. However, at the corporate level, the improvement is only about 5%. Alexander sees this gap as the next frontier, not whether AI can help one project manager draft a charter, but how organizations translate individual gains into enterprise-wide transformation.

The scaling problem is less about technology and more about structures and habits that haven’t caught up. Which brings us back to where this conversation started: the system matters more than the individual. Whether it’s planning a mega-project or deploying AI across a corporation, the leaders who build better systems are the ones whose work endures.

Tags: Project Performance | Project Management | Risk Management | Complexity | Leadership

Related Insights

You May Also Like